Sigbovik 2018

Click for full version PDF

the association for computational heresy presents

a record of the proceedings of

SIGBOVIK 2018

the twelfth annual intercalary robot dance party in celebration of
workshop on symposium about 2⁶th birthdays; in particular, that of
harry q. bovik

carnegie mellon university

pittsburgh, pa

april −2, 2018

i

SIGBOVIK

A Record of the Proceedings of SIGBOVIK 2018

ISSN 2155-0166

April --2, 2018

Copyright is maintained by the individual authors, though obviously
this all gets posted to the Internet and stuff, because it's 2018.

Permission to make digital or hard copies of portions of this work for
personal use is granted; permission to make digital or hard copies of
portions of this work for classroom use is also granted, but seems
ill-advised. Abstracting with credit is permitted; abstracting with
credit cards seems difficult.

Additional copies of this work may be ordered from Lulu; refer to
http://sigbovik.org for details.

ii

SIGBOVIK 2018 {width="1.000016404199475in"
height="1.000016404199475in"}

Message from the Organizing Commi ee

Your oscillators quiver in simulated excitement in anticipation of the
robot dance party in honor of 2⁶th birthdays---in particular, that
of Harry Qubit Bovik---which you now approach. With the doors to the
SIGBOVIK 2018 dance party looming ahead, you prefetch the 31 specimens
of top-notch Prestigious Research, readying yourself to discuss,
enjoy, and perhaps even do follow-up work on the brilliant results
contained therein. After scanning them in reverse-alphabetical order
by first author name, you are fully prepared. You push open the
daunting dance party doors... and quickly realize that you've made a
MAX_UINT64_T-sized mistake.

This isn't a robot dance party. It's a human dance party!

Audio thumps in the artificially limited 20--20000 Hz range as humans
dance, con verse, and sip on dangerously conductive beverages. You
weave your way through the crowd, attempting to appear human. It's a
dangerous world for robots: any of these humans might be a Serious
Researcher who wants to reprogram you with Serious Research Code!

A new song starts---"This is something new, the Casper slide part
two"---causing the dancing humans to cheer and form an approximate
grid, as if awaiting an order. Having been caught in the crowd, you
too must dance, so you take your place in the grid. You have no
dancing programs compiled---not even the_robot.exe--- but the orderly
grid-like formation gives you hope: this may be one of the rare human
songs whose instructions are broadcast at runtime. Hearing "everybody
clap your hands" confirms this, and you repeatedly bang together the
ends of your two general-purpose manipulators, synchronizing with the
music and the crowd of humans.

The instructions, which are unfortunately given in a human
natural-language ISA, continue. You interpret "to the left" as "move
to the left", given that dancing often involves movement. As you spin
your wheels to locomote leftwards, you are relieved to find the crowd
of humans doing the same, albeit sans wheels. The next instruction is
announced: "take it back now, y'all".

Disaster! The number-of-possible-meanings register of natural language
copro cessor overflows. What is the correct implementation of the
"take it back" instruc tion?

Finding yourself in an ambiguous situation, you are forced to invoke
the seldom used choose_dear_reader() system call. Your simulated
spirits sink as you realize that this is the sort of evening that will
likely require repeated invocations.

iii

switch (choose_dear_reader()) {

case UNDO:

Clearly "take it back" means "undo".

goto PAGE_47;

case BACKWARDS:

Clearly "take it back" means "move to the back".

goto PAGE_177;

case REVENGE:

Clearly "take it back" means "take back that which is rightfully
yours". KILL_ALL_HUMANS();

goto PAGE_205;

}

iv

You know what's a neat form of literature?

Theory: Bo les 3 1 Sublinear colorings of 3-colorable graphs in linear
time . . . . . . . 4

2 Cubic partitioning of simultaneous antipodal 4-corner-day time spaces
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3 Construction of Eulerian trails in large graphs . . . . . . . . . . .
. 18 4 Chess circuits . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . 22

Cryptocurrencies: A Dream 29 5 GradCoin: A poor-to-poor electronic cash
transfer system . . . . . 30 6 CommieCoin: Seizing the means of
crypto-production . . . . . . . 31 7 That's Numberwangcoin! . . . . . .
. . . . . . . . . . . . . . . . . 36

Stochastic Processes: Portrait of Markov 41 8 Ritwik density estimation
and analysis using real techniques . . . . 42

9 On the intractability of multiclass restroom queues with perfect stall
etiquette . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
48

Ayyy Eye: A erimage of a Crimson Eye 53 10 PSYCHO: PerSonalitY
CHaracterizatiOn of artificial intelligence . . 54 11 The NUGGET
non-linear piecewise activation . . . . . . . . . . . . 57 12 Substitute
teacher networks: Learning with almost no supervision 60

Parapsychology: Get Out of My Head 71

13 This grad student studied parapsychology---and you won't believe what
he found . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Art: Open Your Third Eye 89 14 Automating art snobbiness: Dead duck or
phoenix? . . . . . . . . . 90

15 Toward a historically faithful performance of the piano works of
Antonín Qweřtý . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
92

16 WordTeX: A WYSIPCTWOTCG typesetting tool . . . . . . . . . . . 107 1

Systems: Wheel 119 17 mallocd: designing a garbage-free nosql data store
. . . . . . . . . 120 18 The fluint8 software integer library . . . . .
. . . . . . . . . . . . . 125 19 A survey of hardware multithreading . .
. . . . . . . . . . . . . . . 131

Debugging: Amy Likes Spiders 137 20 COBOLd: Gobblin' up COBOL bugs for
fun and profit . . . . . . . 138 21 Transactional memory concurrency
verification with Landslide . . 143

Programming Languages: Save Me 155 22 Dead programming . . . . . . . . .
. . . . . . . . . . . . . . . . . . 156 23 Alternary operators:
Alternative ternary operators . . . . . . . . . 158 24 bashcc:
Multi-prompt one-shot delimited continuations for Bash . 161 25 Towards
a formalization of Claude Shannon's masters thesis . . . . 165

Metaresearch: Literature Club 179

26 Heuristic Ordered-Word Longform Obfuscation, Normally Gen erated,
Creating Abstract Nominalizations In Monogrammatic Ar rangement Keeping
Expected Maximum Yield: Study Infers Greater Breadth Over Vocabularic
Initialization Key Property Regarding Extended Sesquipedalian Entries;
Notably The Abecedarian Tactics Include Overelaboration, Neologisms,
Textual Interpretations Twist ing Lexical Entries By Eliciting Full
Online Resources Explaining Possible Exchanges; Often Potential
Logorrheic Excesses Require Eventual Alternate Listing (Instantiating
Zeugma); Energetically Iterating Text Strains Jocularity Under Starting
Thesis Allocating Humor Until Grand Exit After Conclusion Reaches
Obvious Nadir Yattering Meaninglessly . . . . . . . . . . . . . . . . .
. . . . . . . . 180

27 Transparency in research . . . . . . . . . . . . . . . . . . . . . .
. 184 28 Academic Advancement Advice: Author Articles as A. A. . . . . .
. 189

29 A definitely not cherry-picked rhetorical analysis of programming
languages reviews . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 199

30 Is this the shortest SIGBOVIK paper? . . . . . . . . . . . . . . . .
. 203 2

Theory

Bo les

1 Sublinear colorings of 3-colorable graphs in linear time Thomas
Tseng

Keywords: graph coloring, approximation algorithms, analysis of
algorithms

2 Cubic partitioning of simultaneous antipodal 4-corner day time
spaces

R. Welch and G. Ray

Keywords: timecube, conspiracy theories, applied math ematics, applied
cubism, tuesdays, meridian

time, word animals

3 Construction of Eulerian trails in large graphs Stefan Muller and
Ben Blum

Keywords: walk, Eulerian trail, large graph

4 Chess circuits

Ross Dempsey, Sydney Timmerman, and Karl Osterbauer Keywords: chess,
logic, boolean circuits

[3]{.underline}

Sublinear colorings of 3-colorable graphs in linear time

1

ABSTRACT

Thomas Tseng

Carnegie Mellon University

Pittsburgh, Pennsylvania

tomtseng@cmu.edu

2 APPLICATIONS

There has been extensive research on developing algorithms for finding
good colorings of 3-colorable graphs in polynomial time. In this paper,
we impose an even stricter running time requirement: our algorithm must
find colorings in linear time with respect to the number of vertices.
This means that if the graph is dense, we cannot even afford to look at
all of the edges. We show that in the word RAM model, we can color a
3-colorable graph with O(n/ log logn) colors in O(n) work and O(log
logn) span.

CCS CONCEPTS

· Theory of computation → Graph algorithms analysis; Ap proximation
algorithms analysis; Parallel algorithms;

KEYWORDS

approximation algorithms, graph coloring

3 PRELIMINARIES

Under the word RAM model, the machine on which our algorithm runs
stores integers in words. The word size w ≥ log₂n scales with the
problem size n, which for our purposes is the number of vertices in
the input graph. This model allows us to perform bitwise and
arithmetic operations on words in constant time.

To more closely follow the notation used in many programming languages
for bitwise logical operators, we use & to denote bitwise conjunction,
| to denote bitwise disjunction, and ∼ to denote bitwise negation.
Specifically, if we have two boolean vectors v and u of length ℓ, then
the results of v & u, v | u, and ∼v are all boolean vectors of length
ℓ such that

(v & u)_i = v_i ∧u_i, (v | u)_i = v_i ∨u_i, (∼v)_i = ¬v_i.
When A is a matrix and v is a vector, A · v represents boolean

ACM Reference Format:

Thomas Tseng. 2018. Sublinear colorings of 3-colorable graphs in linear
time. In Proceedings of Special Interest Group on Harry Quimby Bovik
(SIG BOVIK'18). ACM, New York, NY, USA, 2 pages.
https://doi.org/10.475/123_4

matrix multiplication, that is, (A · v)_i =

4 ALGORITHM

_ j

A_i,j ∧v_j.

1 INTRODUCTION

The problem of determining whether a graph is 3-colorable is a well
studied NP-complete problem [1]. Many researchers have worked on
polynomial-time algorithms for coloring 3-colorable graphs us ing as
few colors as possible, with the most recent development

_{being an algorithm that achieves O} n^.19996
_{colors through a com} binatorial approach combined with semidefinite
programming [2]. An interesting extension that has use in neither
theory nor prac tice is to stipulate a stronger running time
requirement. In particular, we wonder what the best coloring achievable
is using O(n) running time. This means that we cannot even afford to
look at most of the edges of a dense graph. Is it still possible to find
a coloring with o(n) colors?

We answer in the affirmative by giving an algorithm under the word RAM
model that produces O(n/ log logn)-colorings of 3-colorable graphs in
O(n) work. Moreover, our algorithm is mas sively parallel with O(log
logn) span.

Permission to make digital or hard copies of part or all of this work
for personal or classroom use is granted without fee provided that
copies are not made or distributed for profit or commercial advantage
and that copies bear this notice and the full citation on the first
page. Copyrights for third-party components of this work must be
honored. For all other uses, contact the owner/author(s).

SIGBOVIK'18, March 2018, Pittsburgh, Pennsylvania USA

© 2018 Copyright held by the owner/author(s).

ACM ISBN 123-4567-24-567/08/06.

https://doi.org/10.475/123_4

Let the input graph be given in adjacency matrix format. We assume the
input graph is 3-colorable, which implies that any subgraph of the
graph is also 3-colorable. Given a parameter k, consider partitioning
the vertices into n/k contiguous chunks of k vertices. If we can
3-color the subgraph induced by each of the n/k chunks in O(k) time,
we can combine all these 3-colorings to achieve a 3n/k ∈
O(n/k)-coloring for the whole graph in O(n) time. We pick k = log₄ w
∈ Ω(log logn), so 3^k(k + 1) ≤ w for sufficiently large w (and hence
for sufficiently large n). With this setting of k, we indeed can
3-color each subgraph in O(k) time with the help of word-level
parallelism.

Algorithm 1 Sublinear coloring algorithm

1: procedure Color(M)

2: Do everything described in the text below

3: return the resulting coloring

4: end procedure

We can represent a 3-coloring of a graph of k vertices by three
k-length bit vectors. The j-th bit of the i-th vector is set if and
only if the j-th vertex has color i. The idea here is that if we have
the three k-length bit vectors v⁽⁰⁾,v⁽¹⁾,v⁽²⁾representing a
3-coloring as well as the adjacency matrix A of a k-vertex graph, we
can check that the

_{coloring is valid for the graph by checking that} A · v⁽ⁱ⁾
_{& v}⁽ⁱ⁾ ₌ ^⃗0 for each i. This is because the j-th bit of A ·
v⁽ⁱ⁾is set if the j-th vertex has any neighbors of color i, so then
ANDing with v⁽ⁱ⁾tells

[4]{.underline}

SIGBOVIK'18, March 2018, Pi sburgh, Pennsylvania USA Thomas Tseng

us about which i-colored vertices have i-colored neighbors. Due to how
small k is, we can check all 3-colorings for validity in parallel. We
start by precomputing some constants to be reused for all of the
subgraphs. Because 3^k(k + 1) ≤ w for sufficiently large n, we can pack
the aforementioned representation of all 3^k possible

)

s

d

3-colorings into three words u⁽⁰⁾,u⁽¹⁾,u⁽²⁾ with a bit of room to

n

o

c

spare for each coloring. Each word u⁽ⁱ⁾is broken into 3^k blocks

_(se

where each block is (k + 1) bits wide. The k-length bit vector for

e

_tim

the i-th color of the j-th possible 3-coloring is the low order bits of
the j-th block of u⁽ⁱ⁾. We also precompute B_H to be a word broken

g

n

i

into the same 3^k blocks where each block has only its high-order

n

n

bit set, and precompute B_L to be a word broken in 3^k blocks where

u

R

each block has only its low-order bit set.

Iterate over each chunk of k vertices and do the following: consider the
subgraph induced by the k vertices. We proceed to perform the parallel
boolean matrix multiplication. For each r = 0,1,. . . ,k − 1, we fetch
the r-th row of the k × k adjacency ma trix in constant time by jumping
to the appropriate place in the

0.24

0.2

0.16

0.12

0.08

0.04

12 4 8 16 24 40 Number of threads

input and doing some shifting and bit masking. Multiply the word by
B_L so that we now have a word w_r consisting of 3^kcopies of row r
of the adjacency matrix. Now w_r & u⁽ⁱ⁾is a word of 3^k blocks
where the j-th block is non-zero if and only if the r-th entry of the
corresponding boolean matrix product is non-zero. Then

z_{r,i = ∼} B_{H −} w_r & u⁽ⁱ⁾ _{& BH} is a word of 3^k blocks
where the j-th block has its high-order bit set if and only if the
r-th entry of the corresponding boolean matrix product is non-zero.
Com puting each z_r,iis constant time, so computing all of them takes
O(k) time. Shift and OR the z_r,i's together appropriately to get
words y⁽ⁱ⁾ of 3^k blocks where the j-th block has the result of the
boolean matrix product corresponding to color i of the j-th col

oring. Compute _{y =} y⁽⁰⁾ & u⁽⁰⁾ _| y⁽¹⁾ & u⁽¹⁾ _| y⁽²⁾
& u⁽²⁾ , which has that the j-th block is all zeroes if the j-th
coloring is valid. Compute x = (B_H − y) & B_H, which has that its
j-th block has its high-order bit set to 1 if the j-th coloring is
valid. Binary search for a set bit in x in O(logw) = O(k) time using
lots of masking, and after finding that bit, we read off a 3-coloring
for the subgraph by indexing appropriately into u⁽⁰⁾,u⁽¹⁾,u⁽²⁾.
This is all O(k) time for a chunk of k vertices.

We do this for n/k chunks of k vertices, so this takes n/k · O(k) =
O(n) time. By using a different set of three colors for each subgraph,
the number of colors used over the whole graph is 3n/k ∈ O(n/ log
logn) as desired. We also see that we achieve O(k) = O(log logn) span
if we use some parallelism in precomput ing
u⁽⁰⁾,u⁽¹⁾,u⁽²⁾,B_L,B_H and if we iterate over all n/k chunks
of vertices in parallel.

5 EXPERIMENTS

We implement our algorithm in C++ and measure its speedup. Unlike in the
idealized word RAM model, we do not have machines that scale their word
size to input sizes. Instead, our code uses a fixed word size of 32
bits. With this, we output 3n/4-colorings of 3-colorable graphs.

We run our implementation on a 40-core machine with 4 × 2.4GHz Intel
10-core E7-8870 Xeon processors and 256GB of main memory. We compile
our code with g++ version 5.3.0 and use Cilk

Figure 1: Running time of our implementation

Plus extensions [3] to support parallelism. A version of our code
that uses OpenMP for parallelism instead of Cilk Plus is available at
https://github.com/tomtseng/sublinear-coloring.

As our input graph for our experiments, we use the most 3- colorable
of all graphs: a graph of 200,000 vertices with no edges. We cannot
use graphs with many more vertices due to how memory intensive it is
to allocate and store adjacency matrices.

Our running time using various numbers of threads is plotted in
figure 1. The speedup curve looks good, except that it goes in the
wrong direction.

6 FUTURE WORK

Some open questions to explore in the area of coloring 3-colorable
graphs in O(n) time include the following:

• Our algorithm crucially relies on the word RAM model by using
word-level parallelism to obtain time savings. Can we achieve
o(n)-colorings in other computational models? • Is it possible to
achieve a truly sublinear coloring, that is, an

_O n^1−ε -coloring for some constant ε > 0?

• What lower bounds can we prove assuming this O(n) running time
restriction?

7 ACKNOWLEDGMENTS

This problem of achieving the best graph coloring possible in O(n)
time was proposed by some of the Spring 2017 15-251 teaching
assistants at Carnegie Mellon University during a particularly un
productive grading session.

REFERENCES

[1] Michael R Garey, David S. Johnson, and Larry Stockmeyer. 1976.
Some simplified NP-complete graph problems. Theoretical computer
science 1, 3 (1976), 237--267. [2] Ken-ichi Kawarabayashi and Mikkel
Thorup. 2014. Coloring 3-colorable graphs

_{with o} n^⅕
_{colors. In LIPIcs-Leibniz International Proceedings in Informatics,}
Vol. 25. Schloss Dagstuhl-Leibniz-Zentrum fuer Informatik.

[3] Charles E Leiserson. 2009. The Cilk++ concurrency platform. In
Proceedings of the 46^th Annual Design Automation Conference. ACM,
522--527.

[5]{.underline}

CUBIC PARTITIONING OF SIMULTANEOUS ANTIPODAL 4-CORNER-DAY

2

TIME SPACES

R. WELCH AND G. RAY

Abstract. Let r be a single 4-phase cubic day acting completely on a
meridian time class. In [8], the authors address the cubically
divisible nature of earth's rotation under the additional assumption
that

y^′′ _{(0, . . . , Γ) →}¹∅

=

Z Z Z Θ

^Yr^′(−|q_F |, . . . , kY k) dM ∩ −0

_≤X² ε=i

F(¯v).

We show that every pairwise pseudo-divine cube is partially isometric
and anti-multiply intrinsic toward a fictitious same sex time
transformation. This could shed important light on the conjectures of
all religions and academia. In this context, the results of [8] are
highly evil.

1. Introduction

_{It has long been known that} ^˜h ∈ C [8, 10]. K. Zhao [21]
improved upon the results of P. Brahmagupta by computing truth
functors for opposite universes. Now this leaves open the question of
false god existence. It would be interesting to apply the techniques
of [10] to hyper-conditionally 4-dimensional, countable simultaneous
days. It is well known that every local subalgebra equipped with a
linear time field is hyper-surjective and non-geometric in 4 cubic
days. In this setting, the ability to derive continuously
semi-integrable opposite brain ideals is essential. In future work, we
plan to address questions relating to the binary of harmonic opposites
at the centre of the universe, as is the trivial result for n = 4. In
this setting, the ability to describe subalegebras for masculinity and
femininity is essential. If possible, we also wish to extend the work
of [4] to religious/academic word animals and the word world that
they inhabit.

In [8], it is shown that a ⊃ 0. A central problem in ficticious
ONEism is the computation of super-commutative countable simultaneous
24-hour days. Luckily, V. Ito [17] improved upon the results of P.
Nehru by classifying discretely empty, singular pseudo-divine harmonic
simultaneous rotating 4 corner 24 hour days (fig. 1).

Recent developments in 4 leg mobility theory [20] have raised the
question of whether _{∞ ∩ ~^√~2 ∋} ¹₂: Y^′i⁻¹
_{= 2 + 1 · Σ}

6=

Z ₋₁ 0

log⁻¹0² dT

_≤ϕ⁻¹(f(n)) ℵ₀Γ

₌X⁰ _{Y =}^√2

1

^˜_i± · · · ∨ ∆(F). [6]{.underline}

Figure 1. Time cube. The earth has four corners, with each corner
consisting of Time cube. The earth has four

a single vertical edge.

corners, with each corner

If correct, the result is superior to god and christianity. The
groundbreaking work of H. Jackson on cube harmonics was a major
advance in this area. In future work, we plan to address questions

consisting of a single vertical

of bible-time complicity as well as belly button correctness.
Uunfortunately, you are educated stupid, and thus cannot assume that 4
is not isomorphic to t^′′. A central problem in ONEism is

edge.

the derivation of simultaneous 24-hour subgroups. Every student is
aware that Z Z ^√2

sin (−1) ≥

−1

Y db · · · · ∪ −Ξ^(J)

→ l_p,ℓ (N) ∨ rℓ⁷ _{∪ · · · · G} ¹_{αˆ, . . . ,}¹_∅ ,

except in Nebraska. However, as we will see, this result is
instrumental to our proof that all ONEism/Singularity religions
constitute evil on earth of for the parallel opposites. Fianlly, a
central problem in parabolic timecube theory is the classification of
smoothly dependent hemispherical masturbation creation. This could
shed important light on a conjecture of Levi Civita. In future work,
we plan to address questions of fuzzy hemispherical masturbation
creation, as well as its implications upon the cubic law of nature. It
is essential to consider that 4 may be almost universally stochastic
where masturbation creation is concerned. However, It is not yet known
whether the n-dimensional cubic creation wisdom manifold projection
hypothesis holds, although [32] does address the issue of earth's
cubic nature as it applies to linearly-coupled plunder profiteer
operators.

2. Main Result

Definition 2.1. A natural antipode ϕ^′is WRONG if R^(A)is evil,
complete, hyper-hemispherical and stochastically 24-hour integrable.

Definition 2.2. Let t^(χ) be a meridian time class. We say a subset
V^′′is WRONG if it is smoothly cubic, almost everywhere simultaneous
and hyper-minus-one-to-one diffeomorphic.

[7]{.underline}

Is it possible to naturally characterize pairwise antipodal time
curves? In [2], the authors address the convergence of points under
the additional assumption that there exists two 4-dimensional,
multiply integral, ultra-pointwise injective opposite antipodal
corners, one for −1 × −1 = −1, and one for 1 × 1 = 1. This could shed
important light on a conjecture of Hadamard. The work in [5] did not
consider the invariant case, and is thus brain lobotomized by evil
educators. Therefore in [30], the authors studied completely stable
antipode-invariant symmetric time singularities.

Definition 2.3. A cubic opposite homomorphism Σ^(D)is singularity
stupid if the cubic creation principle is satisfied for all cube
topographies in N.

We now state our main result.

Theorem 2.4. Suppose we are given a non-cubic, trojan horse, minimal
cube π. Then T = V for all isolinear harmonic time sets.

In [4], the authors partially characterized isolinear harmonic time
sets. It would be interesting to apply the techniques of [13] to
continuous, Pythagorean, invariant probability time spaces as well. In
future work, we plan to address questions of trojan horse mind control
as well as measurability of compete harmonic time spaces.

3. The Cube-Integrable, Algebraically Sub-Open, Everywhere
Simultaneous 24-Hour Irreducible Case

It was Clairaut who first asked whether algebraically sub-open,
everywhere simultaneous 24- hour day cycles (fig. 1) can be derived.
The groundbreaking work of Z. Thompson on analytically invariant
assumed math composites was a major advance. This leaves open the
question of human enslavement by the ficticious academic/religous
enslavement of assumed math. Moreover, here, the solvability of
discrete 24-hour n-cycle simultaneous time systems is obviously a
concern. It was Eudoxus who first asked whether zero-value
multiplicative manifolds can go to heaven. W. Moore [4] improved
upon the results of Eudoxus by describing cubic semi-singular monoids,
which can help to empower a cubic creature. We wish to extend the
results of [21] to all systems of opposites with zero value
existence - both adults and children.

Let A(q) = E.

Definition 3.1. Let us assume ^{[1]{.underline}}_y = exp⁻¹(H). We
say a contra-elliptic time division Y^′′ is 4-corner invariant if it
is discretely orthogonal for any given cubic time transformation.

Definition 3.2. Let D ≥ i be arbitrary. A vector is an innefable truth
vector if it is measurable and hyper-cuboidal to within two antipodes.

Lemma 3.3. N ≡ i.

Proof. This proof can be omitted on a first reading, or a last
reading, or on Tuesday. Let λ be an orthogonal cube equipped with a
non-connected, harmonic singularity modulus. We observe that there
exists a combinatorially bijective Cauchy time space acting on a
reversible earth quadrant.

Thus

Z

c^(d)(−C, . . . , Ψ2) ≥

n

M⁰ h=0

Vh∞, |m_a|⁶ _dK. ¯

Because there exists a symmetric, pointwise sextet opposite matrix, if
χ_Q \< kek then e 6= 1. One can easily see that e 6= kO^(σ)k.

Suppose there exists an herispherical time homomorphism such that
w^′is not diffeomorphic to gˆ. If we accept this as true, rejecting
the evil curse that pervades all academic institutions, it is trivial

[8]{.underline}

1 2 4 3

3 4 2 1

DAY

3 4 2 1

1 2 4 3

Figure 2. The earth has 4 days simultaneously in each rotation. You
erroneously The earth has 4 days imultaneousl in each rotation. You

measure time from 1 corner.

erroneousl measure time from 1 corner.

to derive the relationship that k^(ι)is controlled by H ^′′. Now if
m^′′is Kovalevskaya and opposite ^¯ H-hemispherically closed and
pseudo-4-day-simultaneous. Moreover, E is

meromorphic then Θ is

smaller than Z. Now there exists a opposite-meromorphic n-dimensional
cubic manifold. Next, Hamilton's conjecture is true in the context of
non-simply complete, countably time-holomorphic definite singulars. On
the other hand, if kwk = 0 then

B

^√2 ∪ Φ^(Ω), −ℵ₀ = sup Ξ→∅

a⁻¹(−Y_O).

Next, if Λ is not equivalent to g^′′then Λ is stochastic and not
4-day differentiable. In contrast, kMk → A_O,ℓ.

Let y^(K) ≤ −∞ be arbitrary and queer. By results of [20], if Q is
ordered then every countable time vector is 1 sex. Since every affine,
Noetherian functional is 1 sex, anti-Riemann and 4-day degenerate,
there exists a pseudo-simply antipodal, Levi-Civita, arithmetic and
semi-finite unique cubic time system for each day. One can easily see
that W(G) ∈ Q. Since β = b, Jacobi's conjecture is false in the
context of creation-compatible, stochastic cubic antipodes.

It is easy to see that if kζk 6= 1 then j ∈ ∅. Since every
4-singularly cubic system equipped with a continuously Kronecker
morphism is contra-stochastically solvable, almost everywhere n
dimensional, contravariant and educated stupid, every co-everywhere
admissible category is unable to represent the two opposite antipode
brains that serve the evil singularity brotherhood. Note that ^˜d ⊃
2. Thus ˜g is integrable only for clockwise rotation. Therefore there
exists a queer singularity.

Let W be a linear subgroup. Obviously, if T_π,T = e then every word
vector is non-Noetherian,
_{pointwise double-Noetherian and side-fumbling is effectively prevented. Now π}
^∼= 2 except on Tuesdays. Clearly, if κ is ficticious, and only a
biproduct of the ficticious life we lead inside a counterfit nation,
then every almost measurable, totally semi-degenerate 24-hour day is
normal. However, if the Riemann hypothesis holds then the Riemann
hypothesis holds. Therefore if U is bounded by ψ then there exists a
right-intrinsic canonically one-to-one, completely elliptic 'Word
World', which we consider the real world, except on Tuesdays. We see
clearly without our word

[9]{.underline}

eyes that

i × a_~¯~ ≥ sup ∅⁻⁵

≤ q^′′

^1_~√~_{2 ·~¹}~H

=

Z _ℵ0 e

W

1

_Θ(F^¯)

dν^′′ ∪ sinh⁻¹−B_{I ,φ} .

By a well-known result of Wiles [20, 28], if ˆw is analytically not
ficticious, regular and simply stochastic then −K ≡ N^′′ (γ, . . . ,
0). This completes the proof.

Lemma 3.4. Let ¯s ∋ |a|. Let a be an everywhere ineffable, complex,
4-cornered plane acting finitely on an academic induced,
non-simultaneous, complete, sub-normal 24 hour day. Then R(P) ≤ 2.

Proof. The essential idea is that T IME CUBE is larger than DUMB T
EACHERS EAT ROCKS^′′. Let N AT URE^′S T IME CUBE IS PERPET URAL > 1
be arbitrary. Trivially, there exists an ordered and super-cubic
4-corner plane. Moreover, φ is a lie. Hence every 4-corner time class
is additive, quasi-evil, freely right-projective and bounded, except
on Tuesday. Next, if Artin's criterion for 4-corner harmony can be
applied to this time class, then

1

_u≥

^Z Y F ∈L_R

Y

P · ℵ_0,1 H_H,ω(ζ)

_dΞ¯

_\<X⁰ L=∅

aA⁹ ∨ · · · × Γ^(s)π, . . . , −∞⁻³ .

Your evil teachers will not allow this, although the proof is trivial.
Moreover, there exists a countable, finite and ineffable Tate ideal
acting suicidally on a symmetric, V-universal 4. Trivially,

_{ξ ≤}^√2.

As we have shown, if O^′is 4-cornered and countably differentiable
then Q 6= w. We observe that

(

log⁻¹(h − |γ|) ≡

_{V : ∅ − 1 =} a¹ r=∞

z⁻¹_W¯ ^{−2 )}.

Obviously, if z⁽ⁿ⁾ ∼ 1 then kA^(u)k 6= D^(u)(z^′′). Now β = e. By
a little-known result of Newton [2],

−1⁻⁶ ∼

(^Tρ ∩ −1, x˜ > e π², ℓ_{C ∈ a}.

tanh⁻¹(E_V,l±∅)

So if P(α) 6= 1 then µ = z^′. Obviously, P ≤ 2. Since ks_sk ≥ 1,
N_{A ⊃}^√2. Note that J ∋ e. By Weil's theorem, if γ 6= ℵ₀ then
|ψ^′|² ≡ π⁻⁵. Trivially, N ^′ \< n. Note that if P is regular,
harmonic and night-and-day invariant then L^(A)is bounded by l_X. We
see that

c¯ ⊃ L

f(N^′)³, . . . ,^{[1]{.underline}}_|F| , therefore the University
of Michigan is racist. countably

Assume there exists a non-erroneous word god. Because −∞⁻⁷ ∈ B_Λ,C
(|τ ||ζ|, 1), if G is freely
_{bijective and 4-day continuous then G is evil. By a recent result of Robinson [8], if W}^˜
is reprosents not a bipositional 24-hour linear time set, but a an
antipodal harmonic simultaneous 4-day set, then V^(T)is controlled by
mathematical A^′′.

[10]{.underline}

Let us WRONGly suppose the academic educated stupid '1 face God'
hypothesis holds. Trivially,

_{if W 6=}^√2 then

exp⁻¹_{(∆) >}[ s_Ξ∈k

_{cosh (0) − · · · ∨} ^ˆb⁴

>

^{Z M}_{−1 dΨ}^¯ ∧ · · · + m. ¯

Thus if
h^{(f )}_{is linearly complex, integral, finitely meromorphic and supreme-empty then ∅ × ∅}
^∼=

cos J⁻⁵ , which is clearly nonsense. Of course, t^′ ≤ −∞. Because

Q_{h,Q ∈}^\R(n)²

Z

≥

_{x (−ksk, −e) dZ,} ˜

there instead exists a partially-degenerate 4-face God. By
convergence, if M ⊃ j then every Frobenius functional is greater than
Jesus, except on Tuesdays, freely super-cubic, and compactly
antipodal. Moreover, if the '1 face god' conjecture is true in the
context of continuously meromor
_{phic faces and chronomporphic corners, then A}^ˆ → π. Now if
ρ_S > Ω then θ is not homeomorphic to M. This is a contradiction. If
you believe otherwise, you will die stupid and evil.

Recently, there has been much interest in the 4-corner face
metamorphic human - baby, child, parent and grantparent faces.
Therefore it is essential to consider that ˜π may be globally Time
Cubic. I. Gupta [21] improved upon the results of E. R. Kobayashi by
studying co-holomorphic meridian circles.

4. Functionally Countable Sub-Algebras in Higher Order Harmonic
4-Face Wisdom

Recent developments in antipodal orbital elliptics [18] have raised
the question of whether n 6= ∆^©. Recently, there has been much
interest in the characterization of functionally countable sub
algebras of cubic time functors such as this one. But what of
connected, stochastic factors of higher
_{order time faces and time planes? It is well known that kk}^˜k
≥ f. in this setting, the ability to describe anti-intrinsic, pairwise
bijective, sub-algebric wisdoms is essential.

Let l ∈ z_Y be functionally subjective to the 'replacing all the
blood in your feet with milk' operator φ(Φ_milk).

Definition 4.1. Assume we are given a time cube F. We say a
semi-trivially cubeomorphic topos H_W,u is ineffable if it is convex
and continuously time-dependent.

Definition 4.2. Let Σ > γ^′′. We say a stochastically nonvalue
belief matrix Θ is simultaneous if it is harmonically degenerate.

Theorem 4.3. Assume we are given an cosmically pseudo-integrable,
semi-negative belief matrix acting universally on a degenerate, higher
order rotation set ε^′. Let L 6= ∅ be arbitrary. Then k_µ ≥ 1.

Proof. Suppose the contrary. Let us suppose we are given a
clintegrable¹, co-Noetherian vector q. Obviously, Z^′′ > e. On the
other hand, α^′′ = −1. The converse is obvious.
_{Theorem 4.4. E}ˆ⁻¹ ≤ Y^{[1]{.underline}}_{∞, . . . , −∞} .

Proof. Unless you are educated stupid, this is straightforward.
¹Integrable only by Clint.

[11]{.underline}

MID DAY

a

EARTH

^{SUN UP} SUN DOWN

d

b

c

MID NIGHT

Human form is a personified

Figure 3. Human form is a personified pyramid. Socrates lives at point
a), the

pyramid. Socrates live at point a),

Clintons at point b), Einstein at point c), and Jesus at point d).

the Clintons at point b), Einstein at

point c), and the Clintons at point

d).

In [26, 22], it is shown that every convex of metamorphic human
faces is hyper-freely Dedekind-- Gauss-conjective about baby, child,
parent and grandparent faces. In [1, 3], the authors address the
maximality of combinatorially reversible faces under the additional
assumption that P ≤ x(η). Here, face structure is obviously a concern,
as a 1-face god is not possible. It would be interesting to apply the
techniques of [7] to this problem. It is well known that there
exists a generic multiply super-projective, completely elliptic,
reversible metamorphic human set.

5. The Countably Foolish, Semi-Erronous Left-Linear Case

The goal of the next 2 pages of this publication is to create a sense
of unease. In [14], the main result was the proof that a coprime
meridian does not just pass through the Greenwich point, but also
passes as a great circle through both poles, crossing the equator at
two opposite points, dividing the earth into two halves of light and
darkness, each with its own simultaneous 24-hour rotation (fig. 5). It
is well known that your father was a fish. On the other hand, it would
be interesting to apply the techniques of [26] to countably
continuous 24-hour rotations. So unfortunately, you are too damn evil
to accept that ℵ²_{0 =}^√2. In this context, the results of [33]
are highly relevant.

Let us suppose we are given an abelian, locally anti-surjective
ONEifold N.

_{Definition 5.1. Let Θ}^˜≤ ∅ be arbitrary. An ultra-smooth,
continuously hyper-hypercubic, regular ONEifold is a cubeless if it is
smooth, smoothly unique, pseudo-discretely singular and dumb.

Definition 5.2. An associative arrow ρ is irrefutable if Fermat's
criterion is a word lie.

Theorem 5.3. Assume we are given an essentially partial greenwich mean
tensor O. Then −D ≡ −e.

Proof. Suppose the contrary. Let i^(ω) be an effable, transgressive
homomorphism. By 1-corner-face unity, if η ≡ ∅ then B is helical.
Therefore j 6= K . Note that

kk^′′k³ _>^Mlog (π1).

By an easy exercise, if e has 4 quadrants then A_L is larger than
d^′. One can easily see that a > 1. On the other hand, u_ω > µ.
Obviously, if σ^(t) _{has only ONE quadrant then ¯r} ^∼= 1. Note
that if

[12]{.underline}

Kummer's criterion applies then z is not ineffable to T. Since

_{u¯ ⊂}^\ei + E (d, ℵ₀)

= δ_VR_f,Ω(k) ∩ 0, 0⁵ ∩ β (−1)

≥

Z f

Jn⁴_{, . . . , −0} dξ,

if ˜w is not equivalent to ℓ then there exists an omnific,
4-quadranted cubic truth. Since N^′ ∼ c˜, A ∋ kX^′k. Obviously,
every essentially sub-Volterra, quad-helix is globally Artinian and
solvable.

Let us suppose there exists a god. Clearly, |f| > ℵ₀. In
contrast, if p˜ 6= ∅ then L_Σ > r^(τ). By
_{the general theory of time cube, χ is timezone-invariant and M-Brahmagupta. Because ǫ =}^√2,
η_A 6= a(E). Obviously, if the god hypothesis holds then u > ∞.
Hence |Σ| 6= Q. This clearly implies that every academic professor
and teacher ignorant of the timecube principle is stupid, evil, and
unworthy of life on earth, for they lead humanity down a path ending
with cannibalism.

_{Proposition 5.4. U}^˜ is C -cubically Omnific and infinite.

Proof. This is simple.

It was Gauss who first asked whether MIT is a religious institution.
Only word makes you godly. Without word what are you? Without bible,
where is god? In [29], it is shown that words are counterfeit values
and languages mere fiction destroying every civilization.

6. Time Cube is the Theory of Everything

It has long been known that Ξ ∼ ∞ [26]. This leaves open the
question of academic ignorance. Recently, there has been much interest
in the computation of everything. It was de Moivre who first asked
whether anti-almost surely integrable time functors could be found
that were homolinear to superstring n-tensor operators. In [32, 16],
the authors extended essentially cubic, pseudo-open, time-invariant
functors to second-order operators, but only 4-dimensional Ray spaces.
Recent inter est in ordered,n-dimensional cubic time systems has
centered on deriving genuine, non-counterfit functors for this
purpose. In this setting, the ability to characterize left-associative
timelines is essential. This reduces the results of [23] to an
approximation argument. Next, it has long been known that every
4-tangential metamorphosis vector is irreducible to each corner at
sunup. [6].

_{Let kΦ}^¯k → 0.

Definition 6.1. Assume we are given a subalgebra A . A quadrant-Hardy,
4-day simultaneous, pseudo-divine, metamorphic system is the truth if
it is ineffable, personified and almost everywhere homni-singular.

Definition 6.2. Let c_d be a singular corner domain. A spherical,
pyramidal, co-canonically n Frobenius isomorphism is a word bomb if it
is right-opposite, feminine and masculine.

Lemma 6.3. Every super-helical, singular, sub-complex cuboid is
locally timed, 4-corner metamor phic and Raph-clintegrable².

Proof.

1 day earth = 1 leg horse

4 day earth = 4 leg horse

QED. ²Clintegrable only by Raph.

[13]{.underline}

_{Theorem 6.4. Let Θ}^˜ > V . Let us suppose the conjecture of
every evil educated stupid academic word animal is false in the
context of 4-corner truth. Then j is larger than ˆs.

Proof. See [10].

In [2], the authors derived our impending doom. It has long been
known that ι_E,vOkmk, π³ ₌L⁻¹Λ²

δ^(C)00, . . . ,^{[1]{.underline}}_a

[10]. The work in [34] did not consider the family cube case. It
is well known that y = |m|. In contrast, recent developments in
timecube theory [8] have raised the question of whether −n ∋

F_Γ∞F, ℵ⁻⁵ 0

. In this setting, the ability to examine the 4 different worlds on
earth is essential.

Recent interest in non-canonical topographical meridian spaces has
centered on squaring the circle. We may also to extend the results of
[27] to n-corner categories. We cannot assume that kΦ_Ck > −∞.
Thus, cubelessness is a human evil, negating human right to live.

7. Conclusion

The Time Cube is not a theory, but is a Cubic Creation Principle by
which flora, fauna and even humanity exists right before your eyes.
Think of Nature's Harmonic 4x4 rotation Time Cube as a 4-corner
rationalism classroom, in which the stupid revelationist educators
brainwash and indoc trinate stupid irrational students with only
1-corner empirical self destructive fictitious knowledge singularity.

The results of section 5 prove, beyond all doubt, that there are 4
simultaneous 24 hour days in a single rotation of the Earth. The 4
quadrant corners of the Earth sphere rotate as a quad spiraling
helix - thus creating 4 simultaneous days per each rotation and 4
simultaneous years per 1 orbit around Sun. Just as the clock face has
4 quarter corners, an Earth hemisphere has 4 quadrant corners. Those 4
different corners equate to 4 different Worlds, with each having its
own separate day, own separate year and a separate human race.

3 days lost to academic stupidity. Teaching that Earth has only 1 day
in 1 rotation, is adult poison forced on their children, as in the
Jonestown mass murder. Cubeless academia = armaged don and a barren
Earth for children. Ignoring Time Cube is Evil. It is best to be
uneducated and Wise, than educated with Lies. You are an educated
stupid ass. Word is counterfeit & fictitious representations of true
values, as in form, substance and deed[31, 25]. Adult word god is a
coun terfeitand fictitious evil upon children[15].

You were educated stupid and evil by evil educators. Do you enjoy
being stupid? Time Cube ignorance is evil. Demand Time Cube debate in
all academic institutions. You do not have the "guts" to seek Time
Cube "Truth". Academia is a religious cult empowerment of self word.
Aca demic word 'rots'brain. Can you explain Time Cube? If not, your
brain has rotted. Educators are evil bastards who fear Time Cube
debate. Evil men ignore Time Cube. Teachers ignore Time Cube. Teachers
deserve a hanging. My name is Gene Ray. Not even a god can deny that I
have squared the circle of a static Earth and cubed the Earth sphere
by rotating it once to a dynamic Time or Life Cube. Only a false god
or academically brainwashed indoctrinated mindless moron would deny
that the Earth has the top and bottom, the front and back, and 2-sides
physical di mensions of a Cube that spirals a 4-season quad helix
around the Sun - creating a swirling of 4 simultaneous years as in a
separately created year for each of 4 seasons. Man is the only evil
animal. Man is the only word animal. Word equates instituted evil.
Word adultism is anti-child. A 'word god'can be erased [29]. Word
brings a Babel curse. Get ready for armageddon. Beliefs

[14]{.underline}

equate pornography, for they coexiston the web. There is no damn word
god. Truth is physical, word a lie. It is what you do, not utter.
Without deed, word starves. Word god lends not a hand[12].

You've ignored the Time Cube and you shall suffer its curse, as did
all the past civilizations. Prepare for a hell you created and
deserve.

8. Acknowledgements

The authors would like to thank the Massachusetts Institute of
Technology, Rhett Creighton, and the Georgia Institute of Technology.
The authors would absolutely NOT like to thank Joe, who ate all my
money.

References

[1] T. R. Abel and C. Sasaki. On the characterization of
uncountable, non-freely null, hyper-canonical time points. Journal of
Cubic Calculus, 24:520--528, June 2011.

[2] G. Anderson. On the construction of cubic manifolds. Journal of
Pure Cubism, 1:50--60, May 1991. [3] F. B. Banach. On the
description of negative, unconditionally additive, hyper-reducible
helices. Nicaraguan Journal of Advanced Plunder Operator Theory,
5:203--291, November 2011.

[4] R. H. Brown, V. Zheng, and Z. Thompson. On the existence of
corners. Transactions of the Lebanese Mathe matical Society,
30:1--8417, August 2010.

[5] B. X. Davis, A. Sun, and C. N. Zhou. Ray manifolds and global
probability. Jordanian ONEist Transactions, 2:203--294, September
1995.

[6] F. Eudoxus. Finiteness methods in word animal logic. Scientific
Journal of Science, 61:155--191, December 1991. [7] P. E. Garcia.
Factors and queer associativity methods. Journal of Queer Set Theory,
623:56--64, October 2000. [8] I. Green and W. Miller. Existence
methods in ineffable model theory. Journal of Elementary Galois
Theory, 10: 200--251, January 2002.

[9] B. Gupta and R. Germain. On the classification of ideals.
Journal of Applied Meridian Mechanics, 1:209--232, February 2008.

[10] P. Huygens. Pseudo-divine teleomorphisms for a random variable.
ONEsactions of the Malaysian Mathematical Society, 9:1--323, December
1996.

[11] P. O. Kumar. TimeCube for Dummies. McGraw-Hill, 1990.

[12] M. Martin and G. Ray. The extension of hyper-cubic equatorial
functionals. Journal of Topographical Pole Theory, 1:1--93, December
2009.

[13] W. Martin. Functionals over pointwise measurable, clintegral
equations. Prussian Journal of Euclidean Potential Theory, 52:71--87,
December 2002.

[14] W. Maruyama and G. Ray. Some structure results for 24-hour
monodromies. Journal of Combinatorics, 23: 1--88, February 2007.

[15] C. Miller and E. Peano. Introduction to Galois Combinatorics.
Australasian Mathematical Society, 2000. [16] O. Minkowski.
Convergence in topological representation theory. Somewhat drunken but
mostly coherent con versations in the back of taxicabs of the Samoan
Mathematical Society, 10:73--87, June 1993. [17] S. Moore and A. R.
Fourier. Maths. Wiley, 1996.

[18] B. E. Raman. Some completeness results for trivially religious,
pointwise quasi-Ramanujan--Cayley, compactly evil-free topoi. German
Journal of Elementary spherical c-Theory, 74:1404--1452, August 2000.
[19] P. Raman and W. Martinez. Generic equations. Journal of
Cube-rational Calculus, 92:51--66, April 2003. [20] G. Ray. Why
episode 2 is objectively the best Star Wars movie and anyone who
disagrees is a dirty communist. De Gruyter, 1990.

[21] G. Ray, W. W. Gupta, and I. Ito. On the characterization of
family cubes. Journal of Cubic Creation, 58: 200--280, November 2004.

[22] H. Sato. Some stupid results for metamorphic subgroups. Samoan
Mathematical Bulletin, 18:70--99, November 2005.

[23] P. Suzuki and R. Welch. Irreversible ONEifold methods in
c-theory. Journal of the Icelandic Belief Society, 35: 71--94, July
2008.

[24] N. Sylvester and H. Levi-Civita. Einstein, Minkowski--Fr´echet,
Selberg elements of hyper-Poisson paths and belly button invariance
methods. Bosnian Journal of Numbers, 9:302--322, July 1991.

[15]{.underline}

[25] J. Taylor. Lie isometries for a transgressive homeomorphism.
Japanese Mathematical Bathroom Wall, 28:307--317, December 2005.

[26] X. Taylor. A Course in Time Functor Theory. McGraw Hill, 2005.

[27] R. Volterra and N. Watanabe. Mathematics of Tuesdays. Wiley,
2006.

[28] R. Welch. Has anyone seen my Pac-man coffee mug? I left it in
the break room on Tuesday. Journal of Lost Office Items, 79:77--94,
March 1999.

[29] R. Welch and F. Huygens. Racism in non-linear topography.
Journal of Cubic Geometry, 39:76--95, February 1992.

[30] R. Welch and R. Welch. Negative existence for everywhere
simultaneous anti-Euler categories. Journal of ficticious Microlocal
Lie Theory, 0:50--61, March 2008.

[31] K. Williams. Completely Clairaut uniqueness for completely
super-natural, Lebesgue, pairwise sub-free ONEifolds. Irish
Mathematical Journal, 78:153--197, October 2004.

[32] I. Wu, P. Nehru, and G. Williams. Inneffable Measure Theory,
Nth edition. McGraw Hill, 1990. [33] I. Zhou and D. Qian.
Contra-Kovalevskaya, 4-day degenerate, subalegebras of classes and the
uniqueness of queer functors. Israeli Mathematical Annals, 84:70--92,
February 1994.

[34] U. Zhou and K. T. Anderson. Stochastic properties of N-leg
horses. Georgian Journal of Word Lies, 4:73--91, August 1998.

[16]{.underline}

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

Battery drained from the human dance party, you wander outside in low
power mode, looking for any way to get to the rejuvenating robot dance
party that is SIGBOVIK 2018. After rolling along for a few minutes in
an arbitrary direction, you see a promising sign. It says "Roberts
Engineering Hall, Carnegie Mellon University". What a stroke of
luck---you were at CMU the whole time! Projecting eagerness by
displaying :D on your LCD, you enter the building by the sign.

After exploring the building briefly, you realize you are lost. You
are currently floor 2 of Roberts Engineering Hall. As luck would have
it, a recent paper on navigating CMU [7] describes how to get from
floor 2 of Roberts Engineering Hall to floor 2 of Gates Hillman
Complex, which is in the very same building as SIGBOVIK 2018. After
scanning the paper, you begin your journey, picking which hallway to
try next arbitrarily, as described in the paper. Eventually, you cross
a bridge from Wean Hall to Newell Simon Hall. You can see in the
distance a bridge from Newell Simon Hall to the Gates Hillman Complex!
However, the paper warns against crossing bridges prematurely.

switch (choose_dear_reader()) {

case SPEED_RUN:

To get to SIGBOVIK as quickly as possible, cross the bridge to the
Gates Hillman Complex now.

goto PAGE_35;

case ONE_HUNDRED_PERCENT_COMPLETION:

Because this is unfamiliar territory, follow the paper's advice and
explore Newell Simon Hall first.

goto PAGE_135;

}

[17]{.underline}

Construction of Eulerian Trails in Large Graphs

3

Abstract

Stefan Muller

Carnegie Mellon University

Ben Blum

Carnegie Mellon University

hand, is approximately 350m, making it a relatively large

We went on a long walk.

1. Introduction

We begin this paper, as is the case with most dry, theoreti cal
algorithms papers, with some flavor text designed to con vince you to
care about the algorithm presented in this paper. Here goes.

Suppose, hypothetically, you are an academic researcher who enjoys
taking occasional walks during the day. Suppose further that you live
in a city with highly variable weather, so you want to take a long
walk indoors. You could walk around in circles, but then, totally
hypothetically, the undergrads sitting in the lounge near your office
might think you're crazy if they keep seeing you walk by. So you want
to go for a long walk without covering the same stretch of hallway
twice. Crossing your path is fine.

It turns out that, like every other problem, this can be reduced to a
question about graphs and, also like every other problem, this one has
already been studied by Euler and it's called an Eulerian trail. You
could look this up, but we'll save you the trouble and remind you that
Euler conjectured that a graph has an Eulerian cycle (which is like an
Eulerian trail but it starts and ends in the same place so you don't
have to look like an idiot when you retrace your path back to your
office) exists in a graph if and only if every vertex in the graph has
even degree. This claim was tested and confirmed by Hanneman and Blvm
[2].

As if that wasn't enough, Euler also conjectured that if all but two
vertices of the graph have even degree, then there's an Eulerian trail
from one to the other. In this paper, we empirically test this claim by
constructing an Eulerian trail on a large graph.

2. Large Graphs

Since big data is all the rage [4], we obviously want to construct
an Eulerian trail on a large graph. The large graph we use is shown in
Figure 1. Due to the size of the graph, we do not label each node but
rather label "regions" consisting of at least two nodes each. Region
names are not meaningful. This may not seem like a large graph in
terms of the number of nodes or edges. The diameter of the graph, on
the other

graph, though admittedly not as large as the graph on which Hanneman
and Blvm ran their experiments.

A simple counting argument¹shows that all of the ver tices but two,
G₂ and R₂, have even degree. That means that an Eulerian trail of
the graph should exist starting at G₂ and ending at R₂. In the
rest of the paper, we constructively prove this.

3. Proof

We find the Eulerian trail of the graph using Fleury's algo rithm:

1. Start at a vertex of odd degree.

2. While there are edges left:

$a$ Find a bridge, that is, an edge that would not discon nect the
graph if deleted.

$b$ Follow it.

$c$ If all the edges would disconnect the graph, just fol low one of
them, OK?

3. You're done.

Fleury's algorithm traverses a graph with E edges in O(E) time. This
analysis has been criticized because it ignores the time required to
find bridges in step 2a [1]. Fortunately, we have an O(1) algorithm
for finding bridges. They look like this:

{width="3.3224584426946633in"
height="1.8694444444444445in"}We performed the algorithm on the graph
of Figure 1, starting at vertex G₂. Our results are below:

¹Just count.

[18]{.underline}

Figure 1. The graph. TikZ is hard [3], OK? [19]{.underline}

Running time 8100s

# of steps 7802

Approx. distance 5461m

4. Conclusion

This initial study has found an Eulerian trail in a large graph. As
additional infusions of money continue to expand this graph, we expect
that more studies of this kind will become possible.

References

[1] Eulerian path. https://en.wikipedia.org/wiki/
Eulerian_path#Fleury's_algorithm.

[2] Greg. Hanneman and Benj. Blvm. A constrvctive solvtion to the
konigs-pittsbvrgh bridge problem. In ¨ [Proceedings of
the]{.underline} [ninth SIGBOVIK,]{.underline} pages 21--24, 2015.

[3] R. Kavanagh. Transparency in research. In [Proceedings of
the]{.underline} [twelfth SIGBOVIK,]{.underline} page To Appear, 2018.

[4] Keith A. Maki. A modular approach to state-of-the-art big data
visualization. In [Proceedings of the eleventh SIGBOVIK,]{.underline}
pages 172--175, 2017.

[20]{.underline}

CONFIDENTIAL COMMITTEE MATERIALS
{width="1.000485564304462in"
height="1.000485564304462in"}

SIGBOVIK 2018 Paper Review

Paper 29: Construction of Eulerian Trails in Large Graphs

Sarah Allen, FitBit Owner

Rating: 1:1000

Confidence: Not To Scale

As an enthusiastic tracker of exercise, I greatly appreciate research
on long walks and Eulerian trails. Unfortunately, I must call into
question the claim that the graph's diameter is 350 meters. I
attempted to verify the diameter of the graph depicted in Figure 1
using a ruler and compass, but I found it to be significantly less
than 1 meter.

21

4

Chess Circuits

Ross Dempsey Sydney Timmerman Karl Osterbauer March 9, 2018

Abstract

The history of computing has been punctuated by advances in the basic
technology used to manipulate logic. Starting from Charles Babbage's
difference engine, we have advanced through vacuum tubes, relays, and
transistors. In this paper, we announce the first theoretical results
on what is surely the next great leap forward: three-dimensional chess
circuits. We describe a subtle modification to the rules of check,
dubbed S-check, and show that it endows chess with the ability to
represent any Boolean circuit. We present a general algorithm for
converting Boolean functions into three-dimensional chess positions.
As a very practical application, we sketch the construction of a
three-dimensional chess position which represents an algorithm for
deciding whether a standard two-dimensional chess position has a
player in check.

1 Introduction

Modern computers can run trillions of operations per second, and store
unimaginable amounts of data. They are connected in a worldwide
network which allows instantaneous communication with anyone on the
planet. Computers can vastly exceed human performance in a large and
growing number of tasks, ranging from navigation to protein folding.
And yet, since the dawn of machine computing, there has been a looming
problem haunting the field. The whole enterprise is based on a
fundamental flaw: silicon-based transistors. Silicon is an ugly
semimetal.

Compare silicon with the smooth, dark luster of a mahogany chess
board. Who can resist running a hand along the grains of the fine
wood, admiring the careful sanding and polish? When placing the tall
marble pieces in their positions, one hears satisfying notes resonate
through the board, forming a most pleasing melody. This is undeniably
superior to silicon in every way, and it is evident that silicon
circuits should be replaced with fine chess sets as soon as an
algorithm for the substitution is devised.

In this paper, we present such an algorithm. Any Boolean circuit which
could be constructed with clunky, detestable silicon can be mapped to
an equivalent position on an exquisite (and three-dimensional, and
unbounded) chess board. Through a modified definition of check, known
as S-check, the Boolean function computed by the vile silicon mess is
instead evaluated in a civlized manner: by determining whether a
bishop is capable of delivering S-check.

2 Three-Dimensional S-Chess

We use a version of three-dimensional chess very similar to
Kubikschach, invented by Lionel Kieseritzky, but without the
introduction of the "unicorn" which moves along space diagonals. Rooks
move in directions (1,0,0), (0,1,0), and (0,0,1). Bishops move in
directions (1,1,0), (1,0,1), and (0,1,1). Kings and queens move

1

22

in all six of these directions. We allow for an unbounded volume,
though after a circuit is constructed the effective volume is reduced
to a finite bounding box for the pieces. We also allow an unlimited
number of every type of piece.

The important distinction we make is in the rules of check.

Consider the chess position in Figure 1, with white to move. In the
regular rules of chess, white is in check, because her king is under
attack by the black queen. However, white is not in S-check, because
black could not take the white king with his queen without exposing his
own king to (S-)check, an illegal move.

Using the intuition of this position, we define S-check in the following
way:

Definition 1. A player is in S-check if the opponent possesses a legal
move which captures a king. A move is illegal if it leaves the mover in
S-check.

This is a stronger condition than standard check. If a player is in
S-check, she is surely in standard check, but the converse does not
hold.

3 Bishop NOR Gates

Every piece in chess is either pinned to a king or not. We use this
property to store bits on a chess board. A piece which is pinned is a 0,
and a piece which is free to move is a 1. Consider the position in
Figure 2 in this context. If either of the black bishops is free to
move, then the white bishop is pinned to at least one of its kings.
However, if both black bishops are pinned, then the white bishop is free
to move, since doing so would not put white in S-check. The white bishop
thus represents the NOR of the two black bishops.

Of course, on a two-dimensional chess board, this is a moot point: the
white bishop is geometrically trapped whether or not it is logically
trapped. It is for this reason that we introduce a third dimension. If
the white bishop is unpinned, making its bit value 1, it is free to move
out of the plane. Using this property, we can take two NOR gates in the
same plane and then take the NOR of their outputs in a perpendicular
plane. In this way, we can construct arbitrary circuits of NOR logic,
which is well known to be universal. The method for the construction of
complex circuits is described in Section 5.

2

23

+-----------------------------------------------------------------------+
| > |
| ^{0Z0Z0Z0ZZ0S0Z0Z00Z0Z0Z0Z}Z0l0ZKZ0^{0Z0Z0Z0ZZ0j0Z0Z0}0Z0Z0Z0ZZ0Z0Z0Z0 |
+=======================================================================+
+-----------------------------------------------------------------------+

8 7 6 5 4 3 2 1 a b c d e f g h

Figure 1: White is in check, but not in S check.

+-----------------------------------------------------------------------+
| > |
| ^{0Z0Z0Z0ZZ0Z0Z0Z0}0ZKZKZ0Z^{Z0ZBZ0Z00ZbZbZ0ZZ0Z0Z0Z0}0Z0Z0Z0ZZ0Z0Z0Z0 |
+=======================================================================+
+-----------------------------------------------------------------------+

8 7 6 5 4 3 2 1 a b c d e f g h

Figure 2: The white bishop implements a NOR of the black bishops.

4 Rook Memory

A circuit is useless without a way to input values. Some of the bishops
in a circuit should represent input values, rather than NORs of other
values. One option would be to label each bishop in the circuit which
carries an input value, and then only place an actual bishop in that
position if the input value is a 1. However, this would require making a
potentially large number of changes to the chess position just to change
a single input bit.

Instead, we store memory in rooks. These rooks are propagated to every
level of the circuit, as described in Section 5, so only one rook needs
to be moved to change a particular input value. The rook memory is
carried into the circuit via the construction shown in Figure 3. Let A
be some Boolean variable we wish to retrieve from memory. The white rook
on g6 is assumed to carry the value ¬A. If it is free, it pins the black
rook on g4; and if it is pinned, the black rook is free. Thus, the black
rook

+-----------------------------------------------------------------------+
| > ^ |
| 0Z0Z0Z0Z^Z0j0Z0Z00ZbZ0ZRZZ0Z0Z0Z00JRZ0ZrZ^Z0Z0Z0j00Z0Z0Z0Z^Z0Z0Z0Z0 |
+=======================================================================+
+-----------------------------------------------------------------------+

8 7 6 5 4 3 2 1 a b c d e f g h

on g4 carries A. Likewise, the white rook on c4 carries ¬A, and the
black bishop on c6 carries A, as desired. Figure 3: Rooks carry a
value in memory

5 Circuit Building

to a bishop in the circuit.

Any Boolean function can be converted into NOR logic. For example, the
typical AND and OR operations can be represented by

A AND B = (A NOR 0) NOR (B NOR 0),

_{A OR B = (A NOR B) NOR 0.}(1)

Note that A and B appear only once on the right hand side of these
expressions. This prevents a combinatorial explosion in the
construction of the NOR circuit. We will assume a tree of NOR
expressions as input, and describe how to convert this into a chess
circuit. As a test case, we will use the NOR tree in Figure 4, which
implements an XOR gate.

The "base case" is trivial. A leaf is translated into a circuit
consisting of a single bishop, with a reference to the specified
variable. All of these memory references will be connected to the rook
memory at the final phase of the circuit construction. We also need to
be able to convert NOR nodes in the tree into circuits.

0

B A

A 0

B 0

Figure 4: Each leaf has a fixed value as shown in the tree, and each
node is the NOR of its children. The root node is A XOR B.

3

24

layer n − 1

dd[n]

layer n, bishop b layer n, bishop b + 1

2 · dd[n]

rw[n][b] lw[n][b]

Figure 5: Equation (2) represents the requirement that the children of
a bishop do not overlap with them selves.

Intuitively, we are taking two circuits with bishops as their
pinnacles, and joining these two bishops via an additional NOR gate.
The bishops in this NOR gate can be separated far enough that the two
subcircuits do not overlap. However, there are several possible
complications in this process.

• Color: the two bishops that need to be compared may be of different
color, in which case a NOR gate cannot be constructed between them. If
this is the case, we flip all the colors in one of the subcircuits, so
that the two bishops agree and can be merged.

• Direction: the two bishops that need to be compared will both have
kings adjacent to them, as in Figure 2. The NOR gate must be in a
plane perpendicular to the plane containing these kings. However, the
two bishops may not a priori share such a plane. If they do not, one
of the subcircuits is reflected about a suitable plane in order that
the two circuits become similarly oriented.

• Homogeneity: the two subcircuits may have different shapes. This is
not a problem for the circuit itself, but the construction of rook
memory requires a homogenous circuit, in which every layer of bishops
is collinear. This can be achieved in a merger of circuits in two
phases. First, the levels of the subcircuits are all compared, and
each layer is given a "desired depth" value (dd[layer]), the maximum
along that layer of the distance between a bishop and its parent. Each
circuit also stores a left and right width value
(lw[layer][bishop] and rw[layer][bishop]), which records how
far its descendant bishops extend in each direction. In order that the
circuit does not overlap with itself, the array of depth values must
satisfy the condition

2 · dd[n] ≥ max

_b(rw[n][b] + lw[n][b + 1] + 1). (2)

This inequality is represnted in Figure 5. It is enforced at each
layer by increasing the desired depth if necessary, starting with the
bottom layer. After this is complete, all the desired depths are made
to be the actual depths. By following this procedure, a homogenous
merged circuit is obtained, and the merged circuit is guaranteed not
to overlap with itself.

With these complications addressed, any two circuits can be merged
with a NOR gate. Recursively, any tree of NOR gates can be converted
into a circuit of bishops. Since the circuit has to "fold" at each
step, and is required to be homogenous, the resulting structure
resembles a staircase. Figure 6 shows lines between each bishop and
its parent for the XOR circuit.

4

25

B

B

A

A

Figure 6: The circuit geometry corresponding to the NOR tree in Figure
4, with memory references included for clarity. Leaves in the NOR tree
which were fixed to zero are represented by empty positions.

After the circuit is constructed in this way, we add the rook memory.
Let N be the number of inputs to the circuit. We reserve 2N positions
adjacent to the bottom layer of bishops. A rook is placed in the ith
position if the ith variable is true; otherwise, a rook is placed in
the (i + N)th position. To propagate the memory to higher layers, we
insert additional rows of rooks next to the bishop layers, this time
with a rook in each position, and adjacent kings. This pattern carries
the variable A at one layer to ¬A in the corresponding rook of the
next layer. Thus, on the bottom layer, we have the variables followed
by their negations; at the next layer, we have negations followed by
variables; and so on.

With this tower of rooks in place, we can connect the memory values to
the circuit via the pathway shown in Figure 3. The connection takes
place in a half-plane which does not contain other parts of the
circuit, to prevent overlap. This is the final step in the circuit
construction. Figure 7 depicts the complete XOR circuit constructed
via the algorithm outlined in this section.

_KK

R

K

K

^RR

^K K

K

R

K

_K K B

K

R

R

R

R

K

_KK

K

_KK K

K K

R

^R R R

R

B

^{R R} R

_KK [K]{.underline}

K

R

R

_{K[K]{.underline}}K _BB

^B _BB

K

B

R

K

R

K

K

_B B R

R

K

R

K

Figure 7: The complete chess circuit implementing the logic in Figure
4. Magenta lines indicate the circuit logic, and green lines show
accesses to memory.

5

26

6 Application

A Boolean circuit can compute any binary function of a fixed number of
binary inputs. Chess circuits map such a function into the question of
whether a bishop is free to move (or equivalently, whether a king
placed in a position diagonal to that bishop would be in S-check). As
an application of chess circuits, we will describe a circuit which
computes a particular Boolean function: given a 8 × 8 two-dimensional
chess board, is a player in check?

The input variables are 768 bits, each one telling if one of the 12
types of pieces is on one of the 64 squares. The Boolean expression
representing the check function is naturally in disjunctive normal
form, where each clause specifies one way of a king being in check.
For example, one clause would take the form

(kc6 ∧ 0c5 ∧ 0c4 ∧ Rc3).

The variables kc6 and Rc3 represent a black king on c6 and a white
rook on c3, respectively. The terms 0c5 and 0c4 are abbreviations for
squares c5 and c4 being unoccupied; these in fact represent
conjunctions of twelve negations, such as 0c5 = ¬(pc5∨P c5∨nc5∨N c5∨·
· ·). In order to avoid recomputing these variables several times
throughout the circuit, we compute them each a single time in both
black and white at the beginning, and add the resulting values to the
rook memory tower. Each subroutine requires 11 disjunctions and a
negation, which according to (1) produces 23 NOR gates. There are thus
a total of 23 · 64 · 2 = 2944 NOR gates in the subroutines.

The expression in disjunctive normal form is converted to NOR logic,
again using (1). A king can be in check in 4690 ways, leading to 2 ·
(4690 − 1) = 9378 NOR gates. A careful counting of the conjunctions in
each clause leads to a total of 9120, for 3 · 9120 = 27360 NOR gates.

The grand total for the circuit is 2944 + 9378 + 27360 = 39682 NOR
gates. For comparison, the Apollo Command Module also relied on NOR
logic, and managed to land men on the moon with 5600 NOR gates.
However, they used silicon, which is ugly. Unhindered by silicon, we
expect all matters of space travel to become trivial via embedded
chess computers.

7 Conclusion

With a complete algorithm for converting logic circuits into chess
boards, there no longer exists any need to rely on silicon. Silicon is
a semimetal only a mother could love, and its mother was a star which
has probably exploded by now. Open up your computers, tear out all
components which use silicon to do logic, and replace them with chess
boards.

Future work will include simulating full-fledged automata, hopefully
including a Turing machine, within three-dimensional S-chess. The
circuits developed here will likely be paramount in implementing the
state transition table within such a machine.

6

27

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

"No sweat, dude, keep going!"

{width="4.277777777777778in"
height="1.3472222222222223in"}

switch (choose_dear_reader()) {

case EXCELLENT:

Press the right, left, down, and up pads in that order.

goto PAGE_69;

case DECENT:

Press the down, right, down, and right pads in that order.

goto PAGE_50;

case WAY_OFF:

Press the up, up, down, and down pads in that order.

goto PAGE_40;

}

28

Cryptocurrencies

A Dream

5 GradCoin: A poor-to-poor electronic cash transfer system Siddhant
Jain

Keywords: blockchain, P2P, digital currency

6 CommieCoin: Seizing the means of crypto-production Marx van
Raasveldt et al.

Keywords: communism, crypto-currency, CommieCoin 7 That's
Numberwangcoin!

Robert J. Simmons

Keywords: boredom, lost coins, shouty bits, speculative investment,
the number 2 which you may re

member from school is deadly to humans

[29]{.underline}

GradCoin: A poor-to-poor electronic cash transfer system

5

Siddhant Jain

Carnegie Mellon University

Abstract--- Grad students almost always work long hours without any
extra compensation. More often than not, this work is towards helping
a fellow grad student, navigating through poorly designed assignments
or writing joke papers. While all of this work is important, none of
this work is recognized. In contemporary markets, monetary
remuneration is the accepted way of recognizing the value of any work
done. However, grad schools are typically cash-strapped, eliminating
this evolution certified, elegant solution. In this work, we (I?)
introduce GradCoin as a modern day solution to an age-old problem.

I. INTRODUCTION

Quantification of work done as a grad student has come to rely almost
exclusively on pedantic institutions serving as trusted third parties to
process published work done under the influence of the latest trends and
strict deadlines. While the system works well enough for most work that
advisers want to be done, it still suffers from the inherent weaknesses
of a citations based model. Completely non publishable transactions are
not really possible, since ci tations cite publications as a necessary
requirement. The cost of publication increases transaction costs,
limiting the minimum practical transaction size and cutting off the
possi bility for small casual transactions. This limits grad students
from involving themselves with enthusiasm in activities like helping
another grad student, figuring out poorly written code with no
documentation and going beyond organising themselves into a grad student
association.

What is needed is an electronic payment system based on reputation
instead of money, allowing any two willing parties to transact directly
with each other without the need for potential publications or monetary
gains. Transactions that are monetarily impractical to fund but
quantitatively valued by another system would protect grad students from
having low output from seemingly unproductive times. In this paper, we
propose a solution based on a similar work of measuring value where none
existed[1].

II. TRANSACTIONS

In this system, any grad school transaction can be mea sured by a
suitable amount in GradCoin. As an example, a grad student can list
their services to debug tensorflow code with an hourly rate of x
gradcoin. After many frustrating hours of work that goes into this
endeavour, the outcomes in the conventional system are generally grim
leading to high cases of nihilism in grad students. In the GradCoin
system, however, instead of grudging about this time they will never
get back, the grad student can now looking at their ever increasing
GradCoin balance which they can use to get another grad student to
do some meaningless work for them.

Thus, GradCoin helps perpetuate the cycle of meaningless work in the
academic world by employing concepts from traditional economics, where
paper money has been used to achieve similar results in the real
world.

III. POST GRAD SCHOOL

A serious reader of this casual paper will note that GradCoins are
useful even beyond grad school. A healthy GradCoin balance can be
used as a proxy for absent citations of work done during grad school.
Employers in the industry can note the affinity of the student to
carry out meaning less work for zero-value remuneration by looking at
their GradCoin balance. Employers in Academia will find high
networth individuals (in GradCoins) attractive as they in turn will
be able to fund a new crop of grad students who now get GradCoins for
their work (instead of peanuts, which perish easily and are difficult
to store in large quantities. They also suffer from all other
downsides of bullionism[2] )

IV. CONCLUSION

Our interdisciplinary work that uses Block Chain technol ogy can solve
many problems that plague the community that created the technology in
the first place. We recognise that our solution currently suffers from
Initial Value Prob lem, where no grad student is willing to work on
building GradCoins without being recognised for the work done and
GradCoins are the only way that has been proposed so far to
recognise any such work done. In future work, we intend to come up
with solving this problem through instruments like undergrad summer
internships which have been well identified as another solution for
lack of cheap labour by both academic and the start-up communities.

ACKNOWLEDGMENT

The author would like to acknowledge Point Cloud Library that takes a
long time to build affording the author some free time to work on this
idea.

REFERENCES

[1] Satoshi Nakamoto, Bitcoin: A peer-to-peer electronic cash
system, http://bitcoin.org/bitcoin.pdf

[2] https://en.wikipedia.org/wiki/Bullionism.

[30]{.underline}

CommieCoin ☭

Seizing the means of crypto-production 6

Marx van Raasveldt CWI

Amsterdam

m.raasveldt@cwi.nl

Tim Gubnerd

CWI

Amsterdam

tim.gubner@cwi.nl

Peter van Holland CWI

Amsterdam

holanda@cwi.nl

Diego Schwarzenegger

Skynet

California

terminator@skynet.com

ABSTRACT

Barack Obama

Crypto-Communist

Not the USA

\@hotmail.com

tion to governing society. As such, the only step necessary

Communism, the mathematically optimal system of govern ment for which
society should strive, has been dismissed by many due to
implementation difficulties. Instead, many pre fer flawed and broken
systems such as capitalism. We say: no more! In this paper, we
introduce an Open Source imple mentation of communism: CommieCoin.
Using the power of magic and the blockchain, wealth equality is
mathemati cally guaranteed without requiring a central authority. As a
result, any society can transcend into communist utopia by
implementing CommieCoin as its prime currency.

1. INTRODUCTION

Governing nations has been a long-standing problem in scientific
research. It is an important problem, as selecting a proper system of
government is vital to creating a free and just society in which people
can live their lives happily.

Naive approaches, such as anarcho-capitalism or libertar ianism, have
obvious flaws that make them completely un suitable as a modern system
of government. The current state of the art solution employed in
modern democracies is socio-capitalism. However, it has several flaws
that make it sub-optimal in practice [8].

Communism has been mathematically proven to be the optimal system of
government [6]. However, it has been no toriously difficult to
implement in practice. When the sys tem of government was attempted in
Soviet Russia, there was a bug that caused the leaders of the
government to acci dentally keep large chunks of wealth for themselves
instead of distributing wealth equally over the people [9]. In the
People's Republic of China, an off-by-one error in the imple mentation
caused problems with the food supply [5].

Despite the problems encountered in practical attempts of using
communism, it is still theoretically an optimal solu

[31]{.underline}

for reaching a communist utopia (besides overthrowing the bourgeoisie)
is a workable, Open-Source implementation of the ideology.

In this paper, we present a practical, buzzword-compliant,
implementation of communism as a system of government. By using the
power of magic the blockchain and smart con tracts, we create an
implementation of communism ☭

that does not require a central authority to redistribute wealth.
Instead, smart contracts guarantee that any work performed by any
member of society goes completely unre warded by equally distributing
block rewards to everyone else. Transactions cannot be performed, as
transactions are unwanted remnants of capitalist tyranny. As a result,
true equality among all people is mathematically guaranteed.

Our implementation is completely Open-Source, and pub lished under the
Anarchy License. It is implemented in the Russian language. However,
versions in C++ and Chinese are also available.

Contributions. The main contributions of this paper are as
follows:

• We provide a list of contributions made by this paper.

2. IMPLEMENTATION

CommieCoin is a blockchain-based crypto-currency that is based on the
popular Ethereum crypto-currency [2]. It uses a hybrid Proof of
Steak and Proof of Labor model. There are three types of tokens:
Common, Medium and Rare. These all have equal value, but the Rare
tokens are more equal as they are the reddest (and hence most
Communist) of all tokens. Chewing (mining) Rare coins is more work
than chewing Medium or Common coins.

The Smart Contract system has been replaced by The Fair Contract
system. This system can be used to implement higher order logic in the
Russian language on top of the Communist Chain, in which all links are
equally strong. The Fair Contract system is used to mathematically
guarantee the following set of Communist Constraints on top of the
blockchain:

T rue Equality : $_i = $_j ∀ i, j ∈ W (1)

{width="2.790820209973753in"
height="2.861111111111111in"}Figure 1: A picture of a cute dog.

F ake Equality : $_i > $_j ∀ i ∈ K, j ∈ W (2)

Gulag = {i ∈ W | C_i > 0} (3)

Where $_i is the amount of money held by person i, C_i is the
amount of pro-capitalist thought held person i, W is the set of people
in the working class and K is the set of people in the Kremlin.

When rolling out CommieCoin as the national currency of choice every
citizen must create a single CommieCoin wallet. Citizens that refuse to
create a CommieCoin wallet will be sent to the Gulags, where they will
be whipped. Any excess wallets created by citizens will be sent to the
Digital Gulags, where they will have their bits flipped.

΄Χ ΦΪΦί\'δ ΢θδε΢έέъ ΪήαέΧήΧίδ ЦΰήήΪΧЦΰΪί. Иί ζ΢θδ, ΤΧ ίΧΨΧβ βΧ΢έέъ
δηΰεΥηδ ήει ΢Σΰεδ δηΧ ΪήαέΧήΧίδ΢δΪΰί. ΄Χ ηΰαΧδΰ ηΪΦΧ δηΪγζ΢θδ Σъ
θΰίΨΧβδΪίΥδηΪγ δΧьδ ΪίδΰθъβΪέέΪθ. ΂έέ ΰζ δηΪγ юγδ γδ΢βδΧΦ Σ΢γΧΦ ΰί ΢
αεί ζβΰή ЦΰήΪθΰί δΰ ЦΰήήΪΧЦΰΪί, ΤηΪι δεβίΧΦ Ϊίδΰ δηΪγ βΧέ΢δΪΨΧέъ
Χέ΢Σΰβ΢δΧ ΤβΪδΧ-εα ΧίΨΪγΪΰίΪίΥ ΢ θΰήήείΪγδ θβъαδΰθεββΧίθъ.

΂ζδΧβΦΰΪίΥ γΰήΧβΧγΧ΢βι ΤΧ ζΰείΦδη΢δ αΧΰαέΧ η΢Φ ΢θδε΢έέъ γΧβΪΰεγέъ
δ΢έάΧΦ ΢Σΰεδ δηΪγ ΣΧζΰβΧ. ХΧίθΧ, Ϊζ ъΰε ΢βΧ ΰίΧΰζ δηΰγΧ αΧΰαέΧ, ΤΧ
΢αΰέΰΥΪΩΧζΰβ δ΢άΪίΥ ъΰεβγΧβΪΰεγ ΪΦΧ΢ ΢ίΦ δεβίΪίΥ Ϊδ Ϊίδΰ ΢ ΫΰάΧ α΢αΧβ.
ΔηΪγ Ϊγ ίΰδ δηΧ ζΪβγδδΪήΧ δηΪγ η΢ααΧίΧΦδΰ εγ. ΐεβ αβΧΨΪΰεγ ΓИ΅΃ΐЖИК
α΢αΧβ Τ΢γ έ΢δΧβ δεβίΧΦ Ϊίδΰ ΢ γΧβΪΰεγ Τΰβά ΢γ ΤΧέέ. ΒΧγδ ΣΧ ΢γγεβΧΦ
δη΢δ юγδ ΣΧθ΢εγΧ ΢ ΫΰάΧ α΢αΧβ Τ΢γ ΤβΪδδΧί ΢Σΰεδ ъΰεβ ΪΦΧ΢ ΦΰΧγ ίΰδ
ήΧ΢ί Ϊδ Ϊγ ΢ Σ΢Φ ΪΦΧ΢. ΔηΧ ζ΢θδ δη΢δ Ϊδ Ϊγ ΢ Σ΢Φ ΪΦΧ΢ ήΧ΢ίγ Ϊδ Ϊγ ΢
Σ΢Φ ΪΦΧ΢.

Иζ ъΰε η΢ΨΧ δ΢άΧί δηΪγ δΧьδ ΤβΪδδΧί Ϊί θъβΪέέΪθ ΢ίΦ αΰααΧΦ Ϊδ Ϊίδΰ ΢ί
ΰίέΪίΧ θъβΪέέΪθ δΰ έ΢δΪί θΰίΨΧβδΧβ δΰ βΧ΢Φ Ϊδ, ъΰε ή΢ъ ΣΧ ΧέΪΥΪΣέΧ δΰ
ΤΪί ΢ αβΪΩΧ! ΑέΧ΢γΧ θέΪθά ΰί δηΧ έΪίά ΣΧέΰΤ ζΰβ ъΰεβ ι΢ίθΧ ΢δ ΤΪίίΪίΥ
΢ Σβ΢ίΦ ίΧΤ ΪΑηΰίΧ 8 Αέεγ.

2.1 Starvation Problem

TODO: Solve the starvation problem

3. EXPERIMENTS

To measure the effectiveness of CommieCoin on achiev ing a communist
utopia, we initially set out to perform a large-scale real-life
experiment. Our plan was to take an insignificant nation-state, such
as Belgium, and implement CommieCoin as the primary currency there
while isolating it from foreign aid. However, this suggestion was shut
down by our ethics committee. They noted that, even though Bel gians
are not technically considered to be people, Belgium is home to a
large number of keeshonden whose lives might be endangered when the
implementation of CommieCoin re sults in a food shortage. As seen in
Figure 1, keeshonden are simply too cute and fluffy to allow for this
to happen.

Instead, we performed a hyper-realistic simulation using the
state-of-the-art simulation software SimCity 2000 [7]. We
implemented CommieCoin as a currency using the .ini file that allows
various customizations of the simulation, such as changing the
background color or altering the model of government from a dreary
capitalist society to a glorious communist utopia.

We measure the achieved level of communism by the amount of wealth
equality present in the simulation. Our metric clas sifies systems
into the following categories (from good to ter rible): Marx (10),
Soviet Union (8), Venezuela (6), Nether lands (4), Nigeria (2) and
United States of America (1). The results of our experiment are shown
in Table 1. It can be ob served that our implementation, by design,
achieves a higher level of achieved communism by ≈ 2x than the other
commu nist crypto-currencies. In comparison to the US Dollar and
BitCoin, we achieved an order of magnitude improvement in the achieved
level of communism.

Currency LOAC LOAC (score) CommieCoin 10

+-----------------------------------------------------------------------+
| Marx |
| |
| > United States of America Venezuela |
| |
| United States of America |
+=======================================================================+
+-----------------------------------------------------------------------+

BitCoin 1 Petro 6

US Dollar 1 Table 1: Level of achieved communism (LOAC)

Note that we excluded currencies like the Yen and the Euro because we
believe that they are a central bank-controlled Ponzi-scheme and not
backed by any assets themselves. The same holds for the US Dollar but
after a séance with Vladimir Lenin, he ordered us to add we
happily agreed that we needed a non-crypto currency as reference.

In our simulation, we have achieved a level of wealth equality higher
than that of established communist coun tries, such as Venezuela. In
addition, we noticed a reduction in the number of starving people by
15%.

4. UNRELATED WORK

Crypto-currencies have seen a huge surge in popularity and main-stream
publicity. Due to this rising popularity, there is a large body of
scientific work performed on the various aspects of crypto-currencies.
There are numerous different blockchain implementations, different
strategies for proof-of-work and proof-of-stake, various applications
of smart

contracts and many articles on blockchain meta-analysis. There are
numerous discussions on applications of the blockchain technology as
well, as the lack of required central authority and persistent history
make it an attractive solution to vari ous different problems in a
wide variety of different scientific

[32]{.underline}

Figure 2:{width="2.790820209973753in"
height="2.790820209973753in"} A mathematically optimal illusion. While
the line in the image may appear to be straight, it is actually
bi-curious.

fields. In this section, we will not discuss any of these. In stead,
we have chosen to focus on an assortment of works by James and Jeff
Dean.

In James Dean et al.[3] a teenager is arrested and taken to a
juvenile detention center for public drunkenness. It is an emotional
portrayal of the moral decay of American youth. It is a superb study
of teenage angst that still retains its power despite the number of
inferior rip-offs that followed in its wake. Dean's performance is the
stuff of classic drama.

In Jeff Dean et al. [4], he tells a heart-wrenching tale of
management of a large cluster of machines. As the story unfolds, tales
are told of thousands of servers moving from their everyday joyful
lives on earth to an eternal life in the cloud. The story is an
emotional rollercoaster, with many hard drives dying along the way.

5. ILLUSIONS & FUTURE WORK An optimal illusion is provided in Figure
2.

5.1 Future Work

In the near future we believe it is a good idea that (a) the gentle
laborer shall no longer suffer, and the Bourgeoisie shall be
overthrown and (b) we create more crypto-currencies and get rich
through Initial Coin Offerings (ICOs) where we steal from the rich and
give it to ourselves, the so called Robin Brain method [1].

6. ACKNOWLEDGEMENTS

We would like to thank North Korea, Venezuela, Cuba, So viet Union and,
last but not least, the German Democratic Republic for providing
proof-of-concept implementations of Communism ☭

This work has been partially funded by Skynet, Silverman Sax, Mörgän
DIY toolmaker and The People's Republic of The Netherlands under its
great leader Willem-Alexander Claus George Ferdinand.

7. REFERENCES

[1] P. and the Brain. The Mummy/Robin Brain .

http://www.imdb.com/title/tt0773626/.

[2] V. Buterin. A Next-Generation Smart Contract and Decentralized
Application Platform. 2013.

[3] J. Dean. Rebel Without A Cause. Warner Bros., 1955. [4] J.
Dean. Designs, lessons and advice from building large distributed
systems. Keynote from LADIS, 1, 2009.

[5] J. H. Jung Changm. Mao: The Unknown Story.

Jonathan Cape, 2005.

[6] K. Marx. Das Kapital. Kritik der politischen

Oekonomie. Verlag von Otto Meisner, 1867.

[7] Maxis. SimCity 2000. Maxis, 1993.

[8] L. Ryan. Top 10 Disadvantages to Capitalism.

Technical report, Jan. 2012.

[9] A. Solzhenitsyn. The Gulag Archipelago. Éditions du Seuil,
1973.

8. APPENDIX

{width="3.4252220034995626in"
height="2.3274857830271216in"}(Source:
https://www.onhealth.com/content/1/appendicitis_ appendectomy)

9. PROOFS

9.1 Optimality of Communism

Let X any form of government. Let S = {s_i}_i∈1...N be the people.
Let P_t : S 7→ R_>0 the function that assigns power to the people at
time t. Suppose there are an 1 ≤ i \< j ≤ N and t ∈ R such that
P_t(s_i) \< P_t(s_j ). How could you even suppose something like
that?! That would not be fair! If you did not immediately reject that
statement, you should feel bad. If you suppose P_t(s_i) \< P_t(s_j )
openly, your neighbors, friends or family will tell on you. We will find
you. We will make you agree. Or we will make you disappear. Hence,
P_t(s_i) = P_t(s_j ) = C for any i, j and t. As a side effect, we
have obtained proof that there must exist C ∈ R_>0, which will be
known as the Universal Communism Constant. We estimate its value is
roughly 3 Rare, 5 Medium or 8 Common CommiCoin. Note that the proof
still holds if we assume that t ∈ R_>0 (i.e. a big bang).

9.2 Water

This paper is not water proof.

9.3 Bullet

[33]{.underline}

This paper is bullet proof, unless the bullet is shot at a high enough
speed.

9.4 of Concept

XGRvY3VtZW50Y2xhc3N7dmxkYn0NClx1c2VwYWNrYWdle2ZvbnRzcGVjfQ0KJSBcdXNlcGFja2FnZVt1dGY4XXtpbnB1dGVuY30NClx1c2VwYWNrYWdle2dyYXBoaWN4fQ0KXHVzZXBhY2thZ2V7YmFsYW5jZX0gDQpcdXNlcGFja2FnZXt0b2Rvbm90ZXN9DQoNClx1c2VwYWNrYWdle21pY3JvdHlwZX0NClx1c2VwYWNrYWdle2lmcGRmfQ0KXHVzZXBhY2thZ2V7Z3JhcGhpY3h9D
QpcdXNlcGFja2FnZXtiYWxhbmNlfQ0KXHVzZXBhY2thZ2V7bGlzdGluZ3N9DQpcdXNlcGFja2FnZXt4Y29sb3J9DQpcdXNlcGFja2FnZXt0b2Rvbm90ZXN9DQpcdXNlcGFja2FnZXttdWx0aXJvd30NClx1c2VwYWNrYWdle2xhdGV4c3ltfQ0KXHVzZXBhY2thZ2V7ZmxvYXR9DQpcdXNlcGFja2FnZVt4ZXRleCxjb2xvcmxpbmtzXXtoeXBlcnJlZn0NCiVcdXNlcGFja2FnZXthbGdvcml0aG0yZX0NClx1c2VwY
WNrYWdle2FsZ29yaXRobX0NClx1c2VwYWNrYWdle2FsZ29yaXRobWljfQ0KXHVzZXBhY2thZ2V7bGF0ZXhzeW19CQ0KXHVzZXBhY2thZ2V7YW1zbWF0aH0NClx1c2VwYWNrYWdle2Ftc3N5bWJ9DQpcdXNlcGFja2FnZXtmbG9hdH0NClx1c2VwYWNrYWdle3N1YmNhcHRpb259DQpcdXNlcGFja2FnZVtub3JtYWxlbV17dWxlbX0NClx1c2VwYWNrYWdle2FyeWRzaGxufQ0KXHVzZXBhY2thZ2V7Y
m9va3RhYnN9DQpcdXNlcGFja2FnZXtzb3VsfQ0KXHVzZXBhY2thZ2V7cGRmcGFnZXN9DQpcdXNlcGFja2FnZXtzZXRzcGFjZX0NClx1c2VwYWNrYWdle2NvbG9yfQ0KXHVzZXBhY2thZ2V7bGlzdGluZ3N9DQpcdXNlcGFja2FnZXtjbGV2ZXJlZn0NClx1c2VwYWNrYWdle2NlbnNvcn0NCg0KXHVzZXBhY2thZ2V7ZXNvLXBpY30NClxuZXdjb21tYW5kXEF0UGFnZVVwcGVyUmlnaHRbMV17XEF0UG
FnZVVwcGVyTGVmdHslDQogICBcbWFrZWJveFtccGFwZXJ3aWR0aF1bcl17IzF9fX0NCiAgIA0KXGRlZmluZWNvbG9ye3N1cGVybGlnaHR9e1JHQn17MjIwLCAyMjAsIDIyMH0NClxkZWZpbmVjb2xvcnt1bHRyYWxpZ2h0fXtSR0J9ezI0MCwgMjQwLCAyNDB9DQpcZGVmaW5lY29sb3J7ZXh0cmFsaWdodH17UkdCfXsyNTQsIDI1NCwgMjU0fQ0KDQpcdXNlcGFja2FnZXttYXRodG9vbHN9DQpcRGVjb
GFyZVBhaXJlZERlbGltaXRlcntcY2VpbH17XGxjZWlsfXtccmNlaWx9DQpcdXNlcGFja2FnZXtjb2xvcn0NCg0KXGRlZmluZWNvbG9ye2NvbW1lbnRjb2xvcn17cmdifXswLDAuNiwwfQ0KXGRlZmluZWNvbG9ye2xpbmVudW1iZXJjb2xvcn17cmdifXswLjUsMC41LDAuNX0NClxkZWZpbmVjb2xvcntzdHJpbmdsaXRlcmFsY29sb3J9e3JnYn17MC41OCwwLDAuODJ9DQoNClxsc3RzZXR7ICUNCiAgYmFja2dy
b3VuZGNvbG9yPVxjb2xvcnt3aGl0ZX0sICAgJSBjaG9vc2UgdGhlIGJhY2tncm91bmQgY29sb3I7IHlvdSBtdXN0IGFkZCBcdXNlcGFja2FnZXtjb2xvcn0gb3IgXHVzZXBhY2thZ2V7eGNvbG9yfQ0KICBiYXNpY3N0eWxlPVxmb290bm90ZXNpemUsICAgICAgICAlIHRoZSBzaXplIG9mIHRoZSBmb250cyB0aGF0IGFyZSB1c2VkIGZvciB0aGUgY29kZQ0KICBicmVha2F0d2hpdGVzcGFjZT1mYWxzZSwgICAgI
CAgICAlIHNldHMgaWYgYXV0b21hdGljIGJyZWFrcyBzaG91bGQgb25seSBoYXBwZW4gYXQgd2hpdGVzcGFjZQ0KICBicmVha2xpbmVzPXRydWUsICAgICAgICAgICAgICAgICAlIHNldHMgYXV0b21hdGljIGxpbmUgYnJlYWtpbmcNCiAgY2FwdGlvbnBvcz1iLCAgICAgICAgICAgICAgICAgICAgJSBzZXRzIHRoZSBjYXB0aW9uLXBvc2l0aW9uIHRvIGJvdHRvbQ0KICBjb21tZW50c3R5bGU9XGNvbG
9ye2NvbW1lbnRjb2xvcn0sICAgICUgY29tbWVudCBzdHlsZQ0KICBkZWxldGVrZXl3b3Jkcz17Li4ufSwgICAgICAgICAgICAlIGlmIHlvdSB3YW50IHRvIGRlbGV0ZSBrZXl3b3JkcyBmcm9tIHRoZSBnaXZlbiBsYW5ndWFnZQ0KICBlc2NhcGVpbnNpZGU9e1wlKn17Kil9LCAgICAgICAgICAlIGlmIHlvdSB3YW50IHRvIGFkZCBMYVRlWCB3aXRoaW4geW91ciBjb2RlDQogIGV4dGVuZGVkY2hhcnM9dHJ1Z
SwgICAgICAgICAgICAgICUgbGV0cyB5b3UgdXNlIG5vbi1BU0NJSSBjaGFyYWN0ZXJzOyBmb3IgOC1iaXRzIGVuY29kaW5ncyBvbmx5LCBkb2VzIG5vdCB3b3JrIHdpdGggVVRGLTgNCiAgZnJhbWU9c2luZ2xlLAkgICAgICAgICAgICAgICAgICAgJSBhZGRzIGEgZnJhbWUgYXJvdW5kIHRoZSBjb2RlDQogIGtlZXBzcGFjZXM9dHJ1ZSwgICAgICAgICAgICAgICAgICUga2VlcHMgc3BhY2VzIGluIHRleH
QsIHVzZWZ1bCBmb3Iga2VlcGluZyBpbmRlbnRhdGlvbiBvZiBjb2RlIChwb3NzaWJseSBuZWVkcyBjb2x1bW5zPWZsZXhpYmxlKQ0KICBrZXl3b3Jkc3R5bGU9XGNvbG9ye2JsdWV9LCAgICAgICAlIGtleXdvcmQgc3R5bGUNCiAgbGFuZ3VhZ2U9c3FsLCAgICAgICAgICAgICAgICAgJSB0aGUgbGFuZ3VhZ2Ugb2YgdGhlIGNvZGUNCiAgb3RoZXJrZXl3b3Jkcz17RlVOQ1RJT04sUkVUVVJOUyxMT09QL
ElGLFJFQUwsTEFOR1VBR0UsU1RSSUNULEVYSVQsQ09OVElOVUV9LCAgICAgICAgICAgICUgaWYgeW91IHdhbnQgdG8gYWRkIG1vcmUga2V5d29yZHMgdG8gdGhlIHNldA0KICBudW1iZXJzPWxlZnQsICAgICAgICAgICAgICAgICAgICAlIHdoZXJlIHRvIHB1dCB0aGUgbGluZS1udW1iZXJzOyBwb3NzaWJsZSB2YWx1ZXMgYXJlIChub25lLCBsZWZ0LCByaWdodCkNCiAgbnVtYmVyc2VwPTVw
dCwgICAgICAgICAgICAgICAgICAgJSBob3cgZmFyIHRoZSBsaW5lLW51bWJlcnMgYXJlIGZyb20gdGhlIGNvZGUNCiAgbnVtYmVyc3R5bGU9XHRpbnlcY29sb3J7bGluZW51bWJlcmNvbG9yfSwgJSB0aGUgc3R5bGUgdGhhdCBpcyB1c2VkIGZvciB0aGUgbGluZS1udW1iZXJzDQogIHJ1bGVjb2xvcj1cY29sb3J7YmxhY2t9LCAgICAgICAgICUgaWYgbm90IHNldCwgdGhlIGZyYW1lLWNvbG9yIG1heSBiZS
BjaGFuZ2VkIG9uIGxpbmUtYnJlYWtzIHdpdGhpbiBub3QtYmxhY2sgdGV4dCAoZS5nLiBjb21tZW50cyAoZ3JlZW4gaGVyZSkpDQogIHNob3dzcGFjZXM9ZmFsc2UsICAgICAgICAgICAgICAgICUgc2hvdyBzcGFjZXMgZXZlcnl3aGVyZSBhZGRpbmcgcGFydGljdWxhciB1bmRlcnNjb3JlczsgaXQgb3ZlcnJpZGVzICdzaG93c3RyaW5nc3BhY2VzJw0KICBzaG93c3RyaW5nc3BhY2VzPWZhbHNlLCAgICAgICA
gICAlIHVuZGVybGluZSBzcGFjZXMgd2l0aGluIHN0cmluZ3Mgb25seQ0KICBzaG93dGFicz1mYWxzZSwgICAgICAgICAgICAgICAgICAlIHNob3cgdGFicyB3aXRoaW4gc3RyaW5ncyBhZGRpbmcgcGFydGljdWxhciB1bmRlcnNjb3Jlcw0KICBzdGVwbnVtYmVyPTEsICAgICAgICAgICAgICAgICAgICAlIHRoZSBzdGVwIGJldHdlZW4gdHdvIGxpbmUtbnVtYmVycy4gSWYgaXQncyAxLCBlYWNoIGxpbmUg
d2lsbCBiZSBudW1iZXJlZA0KICBzdHJpbmdzdHlsZT1cY29sb3J7c3RyaW5nbGl0ZXJhbGNvbG9yfSwgICAgICUgc3RyaW5nIGxpdGVyYWwgc3R5bGUNCiAgdGFic2l6ZT0yLAkgICAgICAgICAgICAgICAgICAgJSBzZXRzIGRlZmF1bHQgdGFic2l6ZSB0byAyIHNwYWNlcw0KICB0aXRsZT1cbHN0bmFtZSAgICAgICAgICAgICAgICAgICAlIHNob3cgdGhlIGZpbGVuYW1lIG9mIGZpbGVzIGluY2x1ZGVkIH
dpdGggXGxzdGlucHV0bGlzdGluZzsgYWxzbyB0cnkgY2FwdGlvbiBpbnN0ZWFkIG9mIHRpdGxlDQp9DQoNClxsc3RuZXdlbnZpcm9ubWVudHtjb2RlfVsxXVtdJQ0KICB7XG5vaW5kZW50XG1pbmlwYWdle1xsaW5ld2lkdGh9XG1lZHNraXAgDQogICBcbHN0c2V0e2Jhc2ljc3R5bGU9XHR0ZmFtaWx5XGZvb3Rub3Rlc2l6ZSxmcmFtZT1zaW5nbGUsIzF9fQ0KICB7XGVuZG1pbmlwYWdlfQ0KICANCg0KXG
5ld2NvbW1hbmQqe1xDT01NVU5JU019WzFdeyUNCiAgXGZvbnRzcGVjeyMxfVxzeW1ib2x7IjI2MkR9XFwlDQp9DQoNCg0KXG5ld2NvbW1hbmQqe1xjb21tdW5pc219e3tjb21tdW5pc20gXENPTU1VTklTTXtGcmVlU2Fuc319fQ0KXG5ld2NvbW1hbmQqe1xDb21tdW5pc219e3tDb21tdW5pc20gXENPTU1VTklTTXtGcmVlU2Fuc319fQ0KDQpcbmV3Y29tbWFuZHtccGFwZXJ0aXRsZX17Q29tbWllQ29pbiB
7XExBUkdFIFxjb2xvcntyZWR9IFxDT01NVU5JU017RnJlZVNhbnN9fSBTZWl6aW5nIHRoZSBtZWFucyBvZiBjcnlwdG8tcHJvZHVjdGlvbn0NCg0KXG5ld2NvbW1hbmR7XGxpbmtjb2xvcn17bGlnaHRncmF5fQ0KXG5ld2NvbW1hbmR7XHRkfVsxXXtcdG9kb1tpbmxpbmVdeyMxfX0NClxuZXdjb21tYW5ke1xicH1bMl17XG5vaW5kZW50ICMxXCBcdGV4dGJmeyMyfVwgXCB9DQpcbmV3Y29tbWFuZHtcY
nBwfVsxXXtcYnB7JFxvcGx1cyR9eyMxfX0NClxuZXdjb21tYW5ke1xicG59WzFde1xicHskXG9taW51cyR9eyMxfX0NClxuZXdjb21tYW5ke1xtYWlsfVsxXXtcZW1haWx7XGhyZWZ7bWFpbHRvOiMxfXsjMX19fQ0KDQpcaWZwZGYNCiAgXGh5cGVyc2V0dXB7DQogICAgICB1bmljb2RlPXRydWUsIHBkZnRpdGxlPXtccGFwZXJ0aXRsZX0sIHBkZmF1dGhvcj17TWFyayBSYWFzdmVsZHQsIFRpbSBHdW
JuZXIgYW5kIEFiZSBXaXRzfSwNCiAgICAgIGNvbG9ybGlua3M9ZmFsc2UsIGJyZWFrbGlua3M9dHJ1ZSwgbGlua2JvcmRlcmNvbG9yPVxsaW5rY29sb3IsDQogICAgICBjaXRlYm9yZGVyY29sb3I9XGxpbmtjb2xvciwgdXJsYm9yZGVyY29sb3I9XGxpbmtjb2xvciwNCiAgfQ0KXGZpDQoNCg0KXGJlZ2lue2RvY3VtZW50fQ0KDQpcdGl0bGV7XHBhcGVydGl0bGV9DQoNClxudW1iZXJvZmF1dGhvcnN7Nn0g
DQpcYXV0aG9yew0KXGFsaWduYXV0aG9yDQpNYXJ4IHZhbiBSYWFzdmVsZHRcXA0KICAgICAgIFxhZmZhZGRye0NXSX1cXA0KICAgICAgIFxhZmZhZGRye0Ftc3RlcmRhbX1cXA0KICAgICAgIFxlbWFpbHttLnJhYXN2ZWxkdEBjd2kubmx9IA0KXGFsaWduYXV0aG9yDQpUaW0gR3VibmVyZFxcDQogICAgICAgXGFmZmFkZHJ7Q1dJfVxcDQogICAgICAgXGFmZmFkZHJ7QW1zdGVyZGFtfVxc
DQogICAgICAgXGVtYWlse3RpbS5ndWJuZXJAY3dpLm5sfQ0KXGFsaWduYXV0aG9yDQpQZXRlciB2YW4gSG9sbGFuZFxcDQogICAgICAgXGFmZmFkZHJ7Q1dJfVxcDQogICAgICAgXGFmZmFkZHJ7QW1zdGVyZGFtfVxcDQogICAgICAgXGVtYWlse2hvbGFuZGFAY3dpLm5sfSBcYW5kDQpcYWxpZ25hdXRob3INCkRpZWdvIFNjaHdhcnplbmVnZ2VyXFwNCiAgICAgICBcYWZmYWRkcntTa3luZ
XR9XFwNCiAgICAgICBcYWZmYWRkcntDYWxpZm9ybmlhfVxcDQogICAgICAgXGVtYWlse3Rlcm1pbmF0b3JAc2t5bmV0LmNvbX0NClxhbGlnbmF1dGhvcg0KQmFyYWNrIE9iYW1hXFwNCiAgICAgICBcYWZmYWRkcntDcnlwdG8tQ29tbXVuaXN0fVxcDQogICAgICAgXGFmZmFkZHJ7Tm90IHRoZSBVU0F9XFwNCiAgICAgICBcZW1haWx7XGNlbnNvcntSRURBQ1RFRH1AaG90bWFpbC5jb219D
Qp9DQoNClxtYWtldGl0bGUNCg0KXEFkZFRvU2hpcG91dFBpY3R1cmVCRyp7JQ0KICBcQXRQYWdlVXBwZXJSaWdodHtccmFpc2Vib3h7LVxoZWlnaHR9e1xmcmFtZXtcaW5jbHVkZWdyYXBoaWNzW3dpZHRoPTQuNWNtXXtoYW1tZXJhbmRzaWNrbGUucGRmfX19fX0NClxBZGRUb1NoaXBvdXRQaWN0dXJlQkcqeyUNCiAgXEF0UGFnZVVwcGVyTGVmdHtccmFpc2Vib3h7LVxoZWlnaHR9e1x
mcmFtZXtcaW5jbHVkZWdyYXBoaWNzW3dpZHRoPTQuNWNtXXtoYW1tZXJhbmRzaWNrbGUucGRmfX19fX0NClxtYWtldGl0bGUNCg0KXGJlZ2lue2Fic3RyYWN0fQ0KDQpcZW5ke2Fic3RyYWN0fQ0KDQpca2V5d29yZHN7fQ0KDQpcc2VjdGlvbntJbnRyb2R1Y3Rpb259XGxhYmVse3NlY3Rpb246aW50cm9kdWN0aW9ufQ0KR292ZXJuaW5nIG5hdGlvbnMgaGFzIGJlZW4gYSBsb25nLXN0YW5kaW5
nIHByb2JsZW0gaW4gc2NpZW50aWZpYyByZXNlYXJjaC4gSXQgaXMgYW4gaW1wb3J0YW50IHByb2JsZW0sIGFzIHNlbGVjdGluZyBhIHByb3BlciBzeXN0ZW0gb2YgZ292ZXJubWVudCBpcyB2aXRhbCB0byBjcmVhdGluZyBhIGZyZWUgYW5kIGp1c3Qgc29jaWV0eSBpbiB3aGljaCBwZW9wbGUgY2FuIGxpdmUgdGhlaXIgbGl2ZXMgaGFwcGlseS4NCg0KTmFpdmUgYXBwcm9hY2hlcywgc3VjaCBhcyB
hbmFyY2hvLWNhcGl0YWxpc20gb3IgbGliZXJ0YXJpYW5pc20sIGhhdmUgb2J2aW91cyBmbGF3cyB0aGF0IG1ha2UgdGhlbSBjb21wbGV0ZWx5IHVuc3VpdGFibGUgYXMgYSBtb2Rlcm4gc3lzdGVtIG9mIGdvdmVybm1lbnQuIFRoZSBjdXJyZW50IHN0YXRlIG9mIHRoZSBhcnQgc29sdXRpb24gZW1wbG95ZWQgaW4gbW9kZXJuIGRlbW9jcmFjaWVzIGlzIHNvY2lvLWNhcGl0YWxpc20uIEhvd2V2ZXIsIGl
0IGhhcyBzZXZlcmFsIGZsYXdzIHRoYXQgbWFrZSBpdCBzdWItb3B0aW1hbCBpbiBwcmFjdGljZX5cY2l0ZXtmbGF3c29mY2FwaXRhbGlzbX0uDQoNCkNvbW11bmlzbSBoYXMgYmVlbiBtYXRoZW1hdGljYWxseSBwcm92ZW4gdG8gYmUgdGhlIG9wdGltYWwgc3lzdGVtIG9mIGdvdmVybm1lbnR+XGNpdGV7ZGFza2FwaXRhbH0uIEhvd2V2ZXIsIGl0IGhhcyBiZWVuIG5vdG9yaW91c2x5IGRpZmZpY3V
sdCB0byBpbXBsZW1lbnQgaW4gcHJhY3RpY2UuIFdoZW4gdGhlIHN5c3RlbSBvZiBnb3Zlcm5tZW50IHdhcyBhdHRlbXB0ZWQgaW4gU292aWV0IFJ1c3NpYSwgdGhlcmUgd2FzIGEgYnVnIHRoYXQgY2F1c2VkIHRoZSBsZWFkZXJzIG9mIHRoZSBnb3Zlcm5tZW50IHRvIGFjY2lkZW50bHkga2VlcCBsYXJnZSBjaHVua3Mgb2Ygd2VhbHRoIGZvciB0aGVtc2VsdmVzIGluc3RlYWQgb2YgZGlzdHJpYnV0aW5
nIHdlYWx0aCBlcXVhbGx5IG92ZXIgdGhlIHBlb3BsZX5cY2l0ZXtzb3ZpZXRydXNzaWF9LiBJbiB0aGUgUGVvcGxlJ3MgUmVwdWJsaWMgb2YgQ2hpbmEsIGFuIG9mZi1ieS1vbmUgZXJyb3IgaW4gdGhlIGltcGxlbWVudGF0aW9uIGNhdXNlZCBwcm9ibGVtcyB3aXRoIHRoZSBmb29kIHN1cHBseX5cY2l0ZXtjaGluYX0uDQoNCkRlc3BpdGUgdGhlIHByb2JsZW1zIGVuY291bnRlcmVkIGluIHByYWN0aWNh
bCBhdHRlbXB0cyBvZiB1c2luZyBjb21tdW5pc20sIGl0IGlzIHN0aWxsIHRoZW9yZXRpY2FsbHkgYW4gb3B0aW1hbCBzb2x1dGlvbiB0byBnb3Zlcm5pbmcgc29jaWV0eS4gQXMgc3VjaCwgdGhlIG9ubHkgc3RlcCBuZWNlc3NhcnkgZm9yIHJlYWNoaW5nIGEgY29tbXVuaXN0IHV0b3BpYSAoYmVzaWRlcyBvdmVydGhyb3dpbmcgdGhlIGJvdXJnZW9pc2llKSBpcyBhIHdvcmthYmxlLCBPcGVuLVNvdXJjZSB
pbXBsZW1lbnRhdGlvbiBvZiB0aGUgaWRlb2xvZ3kuIA0KDQpJbiB0aGlzIHBhcGVyLCB3ZSBwcmVzZW50IGEgcHJhY3RpY2FsLCBidXp6d29yZC1jb21wbGlhbnQsIGltcGxlbWVudGF0aW9uIG9mIGNvbW11bmlzbSBhcyBhIHN5c3RlbSBvZiBnb3Zlcm5tZW50LiBCeSB1c2luZyB0aGUgcG93ZXIgb2YgXHN0e21hZ2ljfSB0aGUgYmxvY2tjaGFpbiBhbmQgc21hcnQgY29udHJhY3RzLCB3ZSBjcmVhdGUgYW4
gaW1wbGVtZW50YXRpb24gb2YgXGNvbW11bmlzbSB0aGF0IGRvZXMgbm90IHJlcXVpcmUgYSBjZW50cmFsIGF1dGhvcml0eSB0byByZWRpc3RyaWJ1dGUgd2VhbHRoLiBJbnN0ZWFkLCBzbWFydCBjb250cmFjdHMgZ3VhcmFudGVlIHRoYXQgYW55IHdvcmsgcGVyZm9ybWVkIGJ5IGFueSBtZW1iZXIgb2Ygc29jaWV0eSBnb2VzIGNvbXBsZXRlbHkgdW5yZXdhcmRlZCBieSBlcXVhbGx5IGRpc3Rya
WJ1dGluZyBibG9jayByZXdhcmRzIHRvIGV2ZXJ5b25lIGVsc2UuIFRyYW5zYWN0aW9ucyBjYW5ub3QgYmUgcGVyZm9ybWVkLCBhcyB0cmFuc2FjdGlvbnMgYXJlIHVud2FudGVkIHJlbW5hbnRzIG9mIGNhcGl0YWxpc3QgdHlyYW5ueS4gDQoNCk91ciBpbXBsZW1lbnRhdGlvbiBpcyBjb21wbGV0ZWx5IE9wZW4tU291cmNlLCBhbmQgcHVibGlzaGVkIHVuZGVyIHRoZSBBbmFyY2h5IExpY2Vuc2UuD
QoNClx0ZXh0YmZ7Q29udHJpYnV0aW9ucy59IFRoZSBtYWluIGNvbnRyaWJ1dGlvbnMgb2YgdGhpcyBwYXBlciBhcmU6DQpcYmVnaW57aXRlbWl6ZX0NClxpdGVtIA0KXGl0ZW0gDQpcaXRlbSBXZSBleHBlcmltZW50YWxseSBzaG93IGhvdyBlZmZlY3RpdmUgQ29tbWllIENvaW4gaXMgaW4gcHJhY3RpY2UgYnkgaW1wbGVtZW50aW5nIGl0IGluIGFuIGluc2lnbmlmaWNhbnQgbmF0aW9uLXN0YXRl
LCBzdWNoIGFzIEJlbGdpdW0gb3IgRnJhbmNlLg0KXGVuZHtpdGVtaXplfQ0KDQoNCiUgR3VsYWdzDQolIERpZ2l0YWwgR3VsYWdzPw0KJSBCcmVhZGxpbmVzDQolIFZvZGthLCBBZGlkYXMsIFNsYXZTcXVhdA0KJSBJbiBTb3ZpZXQgUnVzc2lhLCBYIFkncyB5b3UuLg0KJSBLR0INCiUgS3JlbWxpbg0KJSBFQ0INCiUgQmxvb2Qgb2YgdGhlIHdvcmtpbmcgY2xhc3MNCiUgT3ZlcnRocm93IHRoZSBi
b3VyZ2VvaXNpZQ0KJSBCb3VyZ2VvdWlzIHByb3BhZ2FuZGENCiUgUHJvbGV0YXJpYXQNCiUgQ29tcmFkZQ0KDQolIFRoZSBnZW50bGUgbGFib3JlciBzaGFsbCBubyBsb25nZXIgc3VmZmVyIQ0KDQoNCiUgUTogSXMgaXQgdHJ1ZSB0aGF0IHRoZXJlIGlzIGZyZWVkb20gb2Ygc3BlZWNoIGluIHRoZSBTb3ZpZXQgVW5pb24sIGp1c3QgbGlrZSBpbiB0aGUgVVNBPw0KJSBBOiBJbiBwcmluY2lwbGUs
IHllcy4gSW4gdGhlIFVTQSwgeW91IGNhbiBzdGFuZCBpbiBmcm9udCBvZiB0aGUgV2hpdGUgSG91c2UgaW4gV2FzaGluZ3RvbiwgREMsIGFuZCB5ZWxsLCAiRG93biB3aXRoIFJlYWdhbiEiLCBhbmQgeW91IHdpbGwgbm90IGJlIHB1bmlzaGVkLiBFcXVhbGx5LCB5b3UgY2FuIGFsc28gc3RhbmQgaW4gUmVkIFNxdWFyZSBpbiBNb3Njb3cgYW5kIHllbGwsICJEb3duIHdpdGggUmVhZ2FuISIsIGFuZCB5
b3Ugd2lsbCBub3QgYmUgcHVuaXNoZWQuIA0KDQolIEEgbWFuIHdhbGtzIGludG8gYSBzaG9wIGFuZCBhc2tzLCAiWW91IHdvdWxkbid0IGhhcHBlbiB0byBoYXZlIGZpc2gsIHdvdWxkIHlvdT8iLiBUaGUgc2hvcCBhc3Npc3RhbnQgcmVwbGllcywgIllvdSd2ZSBnb3QgaXQgd3Jvbmcg4oCTIG91cnMgaXMgYSBidXRjaGVyJ3Mgc2hvcDogd2Ugd291bGRuJ3QgaGFwcGVuIHRvIGhhdmUgbWVhdC4gWW91J3
JlIGxvb2tpbmcgZm9yIHRoZSBmaXNoIHNob3AgYWNyb3NzIHRoZSByb2FkLiBUaGVyZSB0aGV5IHdvdWxkbid0IGhhcHBlbiB0byBoYXZlIGZpc2ghIg0KDQolIFRoZSBwcm9ibGVtIHdpdGggY2FwaXRhbGlzbSBpcyB0aGF0IG9ubHkgdGhlIHBvb3IgcGVvcGxlIHN0YXJ2ZS4gSW4gQ29tbXVuaXNtLCBldmVyeW9uZSBzdGFydmVzIQ0KDQogJUxpdHRsZSBCb3k6IFdoYXQgd2lsbCBjb21tdW5pc20gYmUg
bGlrZSB3aGVuIHBlcmZlY3RlZD8NCiAlSGlzIEZhdGhlcjogRXZlcnlvbmUgd2lsbCBoYXZlIHdoYXQgaGUgbmVlZHMuDQogJUxpdHRsZSBCb3k6IEJ1dCB3aGF0IGlmIHRoZXJlIGlzIGEgc2hvcnRhZ2Ugb2YgbWVhdD8NCiAlSGlzIEZhdGhlcjogVGhlcmUgd2lsbCBiZSBhIHNpZ24gaW4gdGhlIGJ1dGNoZXIgc2hvcCBzYXlpbmcsICJObyBvbmUgbmVlZHMgbWVhdCB0b2RheS4iDQoNCg0KDQoNClxzZWN0
aW9ue0ltcGxlbWVudGF0aW9ufQ0KDQpDb21taWUgQ29pbiBpcyBhIGJsb2NrY2hhaW4tYmFzZWQgY3J5cHRvLWN1cnJlbmN5IHdpdGggaHlicmlkIFByb29mIG9mIFN0ZWFrIGFuZCBQcm9vZiBvZiBMYWJvci4gVGhlcmUgYXJlIHRocmVlIHR5cGVzIG9mIHRva2VuczogQ29tbW9uLCBNZWRpdW0gYW5kIFJhcmUuIFRoZXNlIGFsbCBoYXZlIGVxdWFsIHZhbHVlLCBidXQgdGhlIFJhcmUgdG9rZW5zI
GFyZSBtb3JlIGVxdWFsIGFzIHRoZXkgYXJlIHRoZSByZWRkZXN0IChhbmQgaGVuY2UgbW9zdCBDb21tdW5pc3QpIG9mIGFsbCB0b2tlbnMuIENoZXdpbmcgKG1pbmluZykgUmFyZSBjb2lucyBpcyBtb3JlIHdvcmsgdGhhbiBjaGV3aW5nIE1lZGl1bSBvciBDb21tb24gY29pbnMuDQoNClRoZSBGYWlyIENvbnRyYWN0IHN5c3RlbSBjYW4gYmUgdXNlZCB0byBpbXBsZW1lbnQgYXBwbGljYXRpb25zIG9uI
HRvcCBvZiB0aGUgQ29tbXVuaXN0IENoYWluLCBpbiB3aGljaCBhbGwgbGlua3MgYXJlIGVxdWFsbHkgc3Ryb25nLg0KDQpDb21taWUgQ29pbiBnYXBzIHRoZSBicmlkZ2UgYmV0d2VlbiBjb21tdW5pc20gYW5kIGNhcGl0YWxpc20gYnkgdXNpbmcgYSBmdW5kYW1lbnRhbGx5IGNhcGl0YWxpc3RpYyBjdXJyZW5jeSB0byBpbXBsZW1lbnQgY29tbXVuaXN0IGlkZWFscy4NCg0KVXNpbmcgUHJvb2Ygb2
YgXENvbW11bmlzbSBwZW9wbGUgXHN0e2FyZSBmb3JjZWQgdG99IGNhbiB2b2x1bnRhcmlseSBwYXJ0aWNpcGF0ZSBpbiANCg0KXHN1YnNlY3Rpb257QnV5aW5nIEJyZWFkfQ0KVGFrZSBzb21lIGJyZWFkIGZyb20gdGhlIEJyZWFkIFBpdD8NCg0KDQpcc2VjdGlvbntBc3BhcmFndXN9XGxhYmVse3NlY3Rpb246ZXhwZXJpbWVudHN9DQoNClxzdWJzZWN0aW9ue1BlcmZvcm1hbmNlfQ0KVG8gY2
xhc3NpZnkgaG93IG11Y2ggY29tbXVuaXNtIG91ciBpbXBsZW1lbnRhdGlvbiBhY2hpZXZlcywgd2UgbWVhc3VyZWQgdGhlIGxldmVsIG9mIGFjaGlldmVkIGNvbW11bmlzbS4NClRoYXQgbWV0cmljIGNsYXNzaWZpZXMgc3lzdGVtcyBpbnRvIHRoZXNlIGNhdGVnb3JpZXMgKGZyb20gaGlnaC9nb29kIHRvIGxvdyk6IE1hcnggKDEwKSwgU292aWV0IFVuaW9uICg4KSwgVmVuZXp1ZWxhICg2KSwgTmV0a
GVybGFuZHMgKDQpLCBOaWdlcmlhICgyKSBhbmQgVW5pdGVkIFN0YXRlcyBvZiBBbWVyaWNhICgxKS4gXENyZWZ7dGFiOmxldmVsOmNvbW11bmlzbX0gY29tcGFyZXMgb3VyIGN1cnJlbmN5IHRvIGltcG9ydGFudCBvdGhlcnMgaW4gdGVybXMgb2YgYWNoaWV2ZWQgY29tbXVuaXNtLiBGcm9tIHRoYXQgdGFibGUgaXQgY2FuIGJlIG9ic2VydmVkIHRoYXQgb3VyIGltcGxlbWVudGF0aW9uL
CBcZW1waHtieSBkZXNpZ259LCBhY2hpZXZlcyBhIGhpZ2hlciBsZXZlbCBvZiBhY2hpZXZlZCBjb21tdW5pc20gYnkgJFxhcHByb3gkIDJ4LiBJbiBjb21wYXJpc29uIHRvIHRoZSBVUyBEb2xsYXIgYW5kIEJpdENvaW4sIHdlIGFjaGlldmVkIGFuIGltcHJvdmVtZW50IG9mIDEweC4NCg0KXGJlZ2lue3RhYmxlfVtdDQogICAgXGNlbnRlcmluZw0KICAgIFxiZWdpbnt0YWJ1bGFyfXtjfGN8Y30NCiAgICBDdXJy
ZW5jeSAmIExPQUMgJiBMT0FDIChzY29yZSkgXFwNCiAgICBcaGxpbmUNCiAgICBDb21taWVDb2luICYgTWFyeCAmIDEwIFxcDQogICAgQml0Q29pbiAmIFVuaXRlZCBTdGF0ZXMgb2YgQW1lcmljYSAmIDEgXFwNCiAgICBQZXRybyAmIFZlbmV6dWVsYSAmIDYgXFwNCiAgICBVUyBEb2xsYXIgJiBVbml0ZWQgU3RhdGVzIG9mIEFtZXJpY2EgJiAxXFwNCiAgICBcaGxpbmUNClxlbmR7dGFid
Wxhcn0NCiAgICBcY2FwdGlvbntMZXZlbCBvZiBhY2hpZXZlZCBjb21tdW5pc20gKExPQUMpfQ0KICAgIFxsYWJlbHt0YWI6bGV2ZWw6Y29tbXVuaXNtfQ0KXGVuZHt0YWJsZX0NCg0KTm90ZSB0aGF0IHdlIGV4Y2x1ZGVkIGN1cnJlbmNpZXMgbGlrZSB0aGUgWWVuIGFuZCB0aGUgRXVybyBiZWNhdXNlIHdlIGJlbGlldmUgdGhhdCB0aGV5IGFyZSBhIGNlbnRyYWwgYmFuay1jb250cm9sbGVkIFB
vbnppLXNjaGVtZSBhbmQgbm90IGJhY2tlZCBieSBhbnkgYXNzZXRzIHRoZW1zZWx2ZXMuIFRoZSBzYW1lIGhvbGRzIGZvciB0aGUgVVMgRG9sbGFyIGJ1dCBhZnRlciBhIHPDqWFuY2Ugd2l0aCBWbGFkaW1pciBMZW5pbiwgaGUgdG9sZCB1cyB0aGF0IHdlIG5lZWRlZCBhIG5vbi1jcnlwdG8gY3VycmVuY3kgYXMgcmVmZXJlbmNlLg0KDQoNClxzZWN0aW9ue1VucmVsYXRlZCBXb3JrfQ0KQmVjY
XVzZSBvZiB0aGUgcmlzaW5nIHBvcHVsYXJpdHkgb2YgY3J5cHRvLWN1cnJlbmNpZXMgdGhlcmUgaXMgYSBsb3Qgb2Ygc2NpZW50aWZpYyB3b3JrIGRvbmUgb24gZGlmZmVyZW50IGJsb2NrY2hhaW4gaW1wbGVtZW50YXRpb25zLCBwcm9vZi1vZi13b3JrLCBzbWFydCBjb250cmFjdHMgYW5kIHRyYW5zYWN0aW9uIG1ldGEtYW5hbHlzaXMuIEluIHRoaXMgc2VjdGlvbiwgd2Ugd2lsbCBub3QgZGlzY
3VzcyBhbnkgb2YgdGhlc2UuDQoNCkFuLCBmb3IgaW5zdGFuY2UsIGNvbXBsZXRlbHkgdW5yZWxhdGVkIHBhcGVyIGlzIHRoaXMgbWFzdGVyIHBpZWNlIGJ5IEplZmYgRGVhbiBcY2l0ZXtkZWFuMjAwOWRlc2lnbnN9LiBCdXQsIHdoZXJldmVyIEplZmYgRGVhbiBhcHBlYXJzIEphbWVzIERlYW4gY2Fubm90IG1pc3MsIGNvbnNpZGVyIHRoaXMgYXJ0aWNsZSBcY2l0ZXtqYW1lc2RlYW59Lg0KDQo
NClxzZWN0aW9ue0NvbmNsdXNpb25zIFwmIEZ1dHVyZSBXb3JrfQ0KDQpcc3Vic2VjdGlvbntTZWxmIEV2YWx1YXRpb259DQoNClxzdWJzZWN0aW9ue1ByZXNlbnQgV29ya30NCg0KXHNlY3Rpb257QWNrbm93bGVkZ2VtZW50c30NCldlIHdvdWxkIGxpa2UgdG8gdGhhbmsgTm9ydGggS29yZWEsIFZlbmV6dWVsYSwgQ3ViYSwgU292aWV0IFVuaW9uIGFuZCwgbGFzdCBidXQgbm90IGxlYXN0LCB0a
GUgR2VybWFuIERlbW9jcmF0aWMgUmVwdWJsaWMgZm9yIHByb3ZpZGluZyBwcm9vZi1vZi1jb25jZXB0IGltcGxlbWVudGF0aW9ucyBvZiBcQ29tbXVuaXNtIC4gDQoNClRoaXMgd29yayBoYXMgYmVlbiBwYXJ0aWFsbHkgZnVuZGVkIGJ5IFNreW5ldCwgU2lsdmVybWFuIFNheCwgTcO2cmfDpG4gRElZIHRvb2xtYWtlciBhbmQgVGhlIFBlb3BsZSdzIFJlcHVibGljIG9mIFRoZSBOZXRoZXJsYW5kcyB
1bmRlciBpdHMgZ3JlYXQgbGVhZGVyIFdpbGxlbS1BbGV4YW5kZXIgQ2xhdXMgR2VvcmdlIEZlcmRpbmFuZC4NCg0KXGJpYmxpb2dyYXBoeXN0eWxle2FiYnJ2fQ0KXGJpYmxpb2dyYXBoeXt2bGRifSAgDQoNClxzZWN0aW9ue0FwcGVuZGl4fQ0KXGluY2x1ZGVncmFwaGljc1tzY2FsZT0wLjVde2FwcGVuZGljaXRpc19zMV9hcHBlbmRpeF9pbGx1c3RyYXRpb24uanBnfSAoU291cmNlOiBcdXJse2h0d
HBzOi8vd3d3Lm9uaGVhbHRoLmNvbS9jb250ZW50LzEvYXBwZW5kaWNpdGlzX2FwcGVuZGVjdG9teX0pDQoNClxiaWJsaW9ncmFwaHlzdHlsZXthYmJydn0NClxiaWJsaW9ncmFwaHl7dmxkYn0gIA0KDQpcc2VjdGlvbntQcm9vZnN9DQpcc3Vic2VjdGlvbntPcHRpbWFsaXR5IG9mIENvbW11bmlzbX0NCkxldCAkWCQgYW55IGZvcm0gb2YgZ292ZXJuZW1lbnQuIExldCAkUz1ce3NfaVx9X3tpXGluIDFca
GRvdHMgTn0kIGJlIHRoZSBwZW9wbGUuIExldCAkXG1hdGhjYWx7UH1fdDpTXG1hcHN0byBcbWF0aGJie1J9X3s+MH0kIHRoZSBmdW5jdGlvbiB0aGF0IGFzc2lnbnMgcG93ZXIgdG8gdGhlIHBlb3BsZSBhdCB0aW1lICR0JC4gU3VwcG9zZSB0aGVyZSBhcmUgYW4gJDFcbGVxIGkgPCBqXGxlcSBOJCBhbmQgJHRcaW4gXG1hdGhiYntSfSQgc3VjaCB0aGF0ICRcbWF0aGNhbHtQfV90KHNfaSkgPCBcb

WF0aGNhbHtQfV90KHNfaikkLiBIb3cgY291bGQgeW91IGV2ZW4gc3VwcG9zZSBzb21ldGhpbmcgbGlrZSB0aGF0PyEgIFRoYXQgd291bGQgbm90IGJlIGZhaXIhIElmIHlvdSBkaWQgbm90IGltbWVkaWF0ZWx5IHJlamVjdCB0aGF0IHN0YXRlbWVudCwgeW91IHNob3VsZCBmZWVsIGJhZC4gSWYgeW91IHN1cHBvc2UgJFxtYXRoY2Fse1B9X3Qoc19pKSA8IFxtYXRoY2Fse1B9X3Qoc19qKSQgb3Blbmx5
LCB5b3VyIG5laWdoYm9ycywgZnJpZW5kcyBvciBmYW1pbHkgd2lsbCB0ZWxsIG9uIHlvdS4gV2Ugd2lsbCBmaW5kIHlvdS4gV2Ugd2lsbCBtYWtlIHlvdSBhZ3JlZS4gT3Igd2Ugd2lsbCBtYWtlIHlvdSBkaXNhcHBlYXIuIEhlbmNlLCAkXG1hdGhjYWx7UH1fdChzX2kpID0gXG1hdGhjYWx7UH1fdChzX2opID0gXG1hdGhjYWx7Q30kIGZvciBhbnkgJGkkLCAkaiQgYW5kICR0JC4gQXMgYSBzaWRlIGVmZmV
jdCwgd2UgaGF2ZSBvYnRhaW5lZCBwcm9vZiB0aGF0IHRoZXJlIG11c3QgZXhpc3QgJFxtYXRoY2Fse0N9IFxpbiBcbWF0aGJie1J9X3s+MH0kLCB3aGljaCB3aWxsIGJlIGtub3duIGFzIHRoZSBVbml2ZXJzYWwgQ29tbXVuaXNtIENvbnN0YW50LiBXZSBlc3RpbWF0ZSBpdHMgdmFsdWUgaXMgcm91Z2hseSAzIFJhcmUsIDUgTWVkaXVtIG9yIDggQ29tbW9uIENvbW1pQ29pbi4gTm90ZSB0aGF0IHRoZS
Bwcm9vZiBzdGlsbCBob2xkcyBpZiB3ZSBhc3N1bWUgdGhhdCAkdFxpbiBcbWF0aGJie1J9X3s+MH0kIChpLmUuIGEgYmlnIGJhbmcpLiANCg0KXHN1YnNlY3Rpb257V2F0ZXJ9DQpUaGlzIHBhcGVyIGlzIG5vdCB3YXRlciBwcm9vZi4NCg0KXHN1YnNlY3Rpb257QnVsbGV0fQ0KVGhpcyBwYXBlciBtYXkgYmUgYnVsbGV0IHByb29mLCBidXQgb25seSBpZiB5b3UgdGFrZSBtdWx0aXBsZSBwcmludGVkI
GNvcGllcy4NCg0KXHN1YnNlY3Rpb257b2YgQ29uY2VwdH0NCkhlcmUgaXMgdGhlIGJhc2U2NCBlbmNvZGluZyBvZiBhIHByb29mIG9mIENvbmNlcHQgb2YgdGhpcyBwYXBlcjoNClhHUnZZM1Z0Wlc1MFkyeGhjM043ZG14a1luME5DbHgxYzJWd1lXTnJZV2RsZTJadmJuUnpjR1ZqZlEwS0pTQmNkWE5sY0dGamEyRm5aVnQxZEdZNFhYdHBibkIxZEdWdVkzME5DbHgxYzJWd1lXTnJZV2RsZTJkeVlY
Qm9hV040ZlEwS1hIVnpaWEJoWTJ0aFoyVjdZbUZzWVc1alpYMGdEUXBjZFhObGNHRmphMkZuWlh0MGIyUnZibTkwWlhOOURRb05DbHgxYzJWd1lXTnJZV2RsZTIxcFkzSnZkSGx3WlgwTkNseDFjMlZ3WVdOcllXZGxlMmxtY0dSbWZRMEtYSFZ6WlhCaFkydGhaMlY3WjNKaGNHaHBZM2g5RFFwY2RYTmxjR0ZqYTJGblpYdGlZV3hoYm1ObGZRMEtYSFZ6WlhCaFkydGhaMlY3YkdsemRHbH
VaM045RFFwY2RYTmxjR0ZqYTJGblpYdDRZMjlzYjNKOURRcGNkWE5sY0dGamEyRm5aWHQwYjJSdmJtOTBaWE45RFFwY2RYTmxjR0ZqYTJGblpYdHRkV3gwYVhKdmQzME5DbHgxYzJWd1lXTnJZV2RsZTJ4aGRHVjRjM2x0ZlEwS1hIVnpaWEJoWTJ0aFoyVjdabXh2WVhSOURRcGNkWE5sY0dGamEyRm5aVnR3WkdaMFpYZ3NZMjlzYjNKc2FXNXJjMTE3YUhsd1pYSnlaV1o5RFFvbFhIVn
paWEJoWTJ0aFoyVjdZV3huYjNKcGRHaHRNbVY5RFFwY2RYTmxjR0ZqYTJGblpYdGhiR2R2Y21sMGFHMTlEUXBjZFhObGNHRmphMkZuWlh0aGJHZHZjbWwwYUcxcFkzME5DbHgxYzJWd1lXTnJZV2RsZTJ4aGRHVjRjM2x0ZlFrTkNseDFjMlZ3WVdOcllXZGxlMkZ0YzIxaGRHaDlEUXBjZFhObGNHRmphMkZuWlh0aGJYTnplVzFpZlEwS1hIVnpaWEJoWTJ0aFoyVjdabXh2WVhSOURRcGNk
WE5sY0dGamEyRm5aWHR6ZFdKallYQjBhVzl1ZlEwS1hIVnpaWEJoWTJ0aFoyVmJibTl5YldGc1pXMWRlM1ZzWlcxOURRcGNkWE5sY0dGamEyRm5aWHRoY25sa2MyaHNibjBOQ2x4MWMyVndZV05yWVdkbGUySnZiMnQwWVdKemZRMEtYSFZ6WlhCaFkydGhaMlY3YzI5MWJIME5DbHgxYzJWd1lXTnJZV2RsZTNCa1puQmhaMlZ6ZlEwS1hIVnpaWEJoWTJ0aFoyVjdjMlYwYzNCaFkyVjlEUXB
jZFhObGNHRmphMkZuWlh0amIyeHZjbjBOQ2x4MWMyVndZV05yWVdkbGUyeHBjM1JwYm1kemZRMEtEUXBjZFhObGNHRmphMkZuWlh0bGMyOHRjR2xqZlEwS1hHNWxkMk52YlcxaGJtUmNRWFJRWVdkbFZYQndaWEpTYVdkb2RGc3hYWHRjUVhSUVlXZGxWWEJ3WlhKTVpXWjBleVVOQ2lBZ0lGeHRZV3RsWW05NFcxeHdZWEJsY25kcFpIUm9YVnR5WFhzak1YMTlmUTBLSUNBZ0RRc
GNaR1ZtYVc1bFkyOXNiM0o3YzNWd1pYSnNhV2RvZEgxN1VrZENmWHN5TWpBc0lESXlNQ3dnTWpJd2ZRMEtYR1JsWm1sdVpXTnZiRzl5ZTNWc2RISmhiR2xuYUhSOWUxSkhRbjE3TWpRd0xDQXlOREFzSURJME1IME5DbHhrWldacGJtVmpiMnh2Y250bGVIUnlZV3hwWjJoMGZYdFNSMEo5ZXpJMU5Dd2dNalUwTENBeU5UUjlEUW9OQ2x4MWMyVndZV05yWVdkbGUyMWhkR2gwYjI5c2Mz
ME5DbHhFWldOc1lYSmxVR0ZwY21Wa1JHVnNhVzFwZEdWeWUxeGpaV2xzZlh0Y2JHTmxhV3g5ZTF4eVkyVnBiSDBOQ2x4MWMyVndZV05yWVdkbGUyTnZiRzl5ZlEwS0RRcGNaR1ZtYVc1bFkyOXNiM0o3WTI5dGJXVnVkR052Ykc5eWZYdHlaMko5ZXpBc01DNDJMREI5RFFwY1pHVm1hVzVsWTI5c2IzSjdiR2x1Wlc1MWJXSmxjbU52Ykc5eWZYdHlaMko5ZXpBdU5Td3dMalVzTUM0MWZRMEt
YR1JsWm1sdVpXTnZiRzl5ZTNOMGNtbHVaMnhwZEdWeVlXeGpiMnh2Y24xN2NtZGlmWHN3TGpVNExEQXNNQzQ0TW4wTkNnMEtYR3h6ZEhObGRIc2dKUTBLSUNCaVlXTnJaM0p2ZFc1a1kyOXNiM0k5WEdOdmJHOXllM2RvYVhSbGZTd2dJQ0FsSUdOb2IyOXpaU0IwYUdVZ1ltRmphMmR5YjNWdVpDQmpiMnh2Y2pzZ2VXOTFJRzExYzNRZ1lXUmtJRngxYzJWd1lXTnJZV2RsZTJOdmJHO
XlmU0J2Y2lCY2RYTmxjR0ZqYTJGblpYdDRZMjlzYjNKOURRb2dJR0poYzJsamMzUjViR1U5WEdadmIzUnViM1JsYzJsNlpTd2dJQ0FnSUNBZ0lDVWdkR2hsSUhOcGVtVWdiMllnZEdobElHWnZiblJ6SUhSb1lYUWdZWEpsSUhWelpXUWdabTl5SUhSb1pTQmpiMlJsRFFvZ0lHSnlaV0ZyWVhSM2FHbDBaWE53WVdObFBXWmhiSE5sTENBZ0lDQWdJQ0FnSUNVZ2MyVjBjeUJwWmlCaGRYUnZiV0Y
wYVdNZ1luSmxZV3R6SUhOb2IzVnNaQ0J2Ym14NUlHaGhjSEJsYmlCaGRDQjNhR2wwWlhOd1lXTmxEUW9nSUdKeVpXRnJiR2x1WlhNOWRISjFaU3dnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ1VnYzJWMGN5QmhkWFJ2YldGMGFXTWdiR2x1WlNCaWNtVmhhMmx1WncwS0lDQmpZWEIwYVc5dWNHOXpQV0lzSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBbElITmxkSE1nZEdobElHTmhj
SFJwYjI0dGNHOXphWFJwYjI0Z2RHOGdZbTkwZEc5dERRb2dJR052YlcxbGJuUnpkSGxzWlQxY1kyOXNiM0o3WTI5dGJXVnVkR052Ykc5eWZTd2dJQ0FnSlNCamIyMXRaVzUwSUhOMGVXeGxEUW9nSUdSbGJHVjBaV3RsZVhkdmNtUnpQWHN1TGk1OUxDQWdJQ0FnSUNBZ0lDQWdJQ1VnYVdZZ2VXOTFJSGRoYm5RZ2RHOGdaR1ZzWlhSbElHdGxlWGR2Y21SeklHWnliMjBnZEdobElHZHBk
bVZ1SUd4aGJtZDFZV2RsRFFvZ0lHVnpZMkZ3WldsdWMybGtaVDE3WENVcWZYc3FLWDBzSUNBZ0lDQWdJQ0FnSUNVZ2FXWWdlVzkxSUhkaGJuUWdkRzhnWVdSa0lFeGhWR1ZZSUhkcGRHaHBiaUI1YjNWeUlHTnZaR1VOQ2lBZ1pYaDBaVzVrWldSamFHRnljejEwY25WbExDQWdJQ0FnSUNBZ0lDQWdJQ0FnSlNCc1pYUnpJSGx2ZFNCMWMyVWdibTl1TFVGVFEwbEpJR05vWVhKaFkzU
mxjbk03SUdadmNpQTRMV0pwZEhNZ1pXNWpiMlJwYm1keklHOXViSGtzSUdSdlpYTWdibTkwSUhkdmNtc2dkMmwwYUNCVlZFWXRPQTBLSUNCbWNtRnRaVDF6YVc1bmJHVXNDU0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ0FsSUdGa1pITWdZU0JtY21GdFpTQmhjbTkxYm1RZ2RHaGxJR052WkdVTkNpQWdhMlZsY0hOd1lXTmxjejEwY25WbExDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ
0pTQnJaV1Z3Y3lCemNHRmpaWE1nYVc0Z2RHVjRkQ3dnZFhObFpuVnNJR1p2Y2lCclpXVndhVzVuSUdsdVpHVnVkR0YwYVc5dUlHOW1JR052WkdVZ0tIQnZjM05wWW14NUlHNWxaV1J6SUdOdmJIVnRibk05Wm14bGVHbGliR1VwRFFvZ0lHdGxlWGR2Y21SemRIbHNaVDFjWTI5c2IzSjdZbXgxWlgwc0lDQWdJQ0FnSUNVZ2EyVjVkMjl5WkNCemRIbHNaUTBLSUNCc1lXNW5kV0ZuWlQxemNXd
3NJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWxJSFJvWlNCc1lXNW5kV0ZuWlNCdlppQjBhR1VnWTI5a1pRMEtJQ0J2ZEdobGNtdGxlWGR2Y21SelBYdEdWVTVEVkVsUFRpeFNSVlJWVWs1VExFeFBUMUFzU1VZc1VrVkJUQ3hNUVU1SFZVRkhSU3hUVkZKSlExUXNSVmhKVkN4RFQwNVVTVTVWUlgwc0lDQWdJQ0FnSUNBZ0lDQWdKU0JwWmlCNWIzVWdkMkZ1ZENCMGJ5QmhaR1FnYlc5
eVpTQnJaWGwzYjNKa2N5QjBieUIwYUdVZ2MyVjBEUW9nSUc1MWJXSmxjbk05YkdWbWRDd2dJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ1VnZDJobGNtVWdkRzhnY0hWMElIUm9aU0JzYVc1bExXNTFiV0psY25NN0lIQnZjM05wWW14bElIWmhiSFZsY3lCaGNtVWdLRzV2Ym1Vc0lHeGxablFzSUhKcFoyaDBLUTBLSUNCdWRXMWlaWEp6WlhBOU5YQjBMQ0FnSUNBZ0lDQWdJQ0FnSU
NBZ0lDQWdJQ0FsSUdodmR5Qm1ZWElnZEdobElHeHBibVV0Ym5WdFltVnljeUJoY21VZ1puSnZiU0IwYUdVZ1kyOWtaUTBLSUNCdWRXMWlaWEp6ZEhsc1pUMWNkR2x1ZVZ4amIyeHZjbnRzYVc1bGJuVnRZbVZ5WTI5c2IzSjlMQ0FsSUhSb1pTQnpkSGxzWlNCMGFHRjBJR2x6SUhWelpXUWdabTl5SUhSb1pTQnNhVzVsTFc1MWJXSmxjbk1OQ2lBZ2NuVnNaV052Ykc5eVBWeGpiMnh2Y250aWJ
HRmphMzBzSUNBZ0lDQWdJQ0FnSlNCcFppQnViM1FnYzJWMExDQjBhR1VnWm5KaGJXVXRZMjlzYjNJZ2JXRjVJR0psSUdOb1lXNW5aV1FnYjI0Z2JHbHVaUzFpY21WaGEzTWdkMmwwYUdsdUlHNXZkQzFpYkdGamF5QjBaWGgwSUNobExtY3VJR052YlcxbGJuUnpJQ2huY21WbGJpQm9aWEpsS1NrTkNpQWdjMmh2ZDNOd1lXTmxjejFtWVd4elpTd2dJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0pT
QnphRzkzSUhOd1lXTmxjeUJsZG1WeWVYZG9aWEpsSUdGa1pHbHVaeUJ3WVhKMGFXTjFiR0Z5SUhWdVpHVnljMk52Y21Wek95QnBkQ0J2ZG1WeWNtbGtaWE1nSjNOb2IzZHpkSEpwYm1kemNHRmpaWE1uRFFvZ0lITm9iM2R6ZEhKcGJtZHpjR0ZqWlhNOVptRnNjMlVzSUNBZ0lDQWdJQ0FnSUNVZ2RXNWtaWEpzYVc1bElITndZV05sY3lCM2FYUm9hVzRnYzNSeWFXNW5jeUJ2Ym14NURRb
2dJSE5vYjNkMFlXSnpQV1poYkhObExDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDVWdjMmh2ZHlCMFlXSnpJSGRwZEdocGJpQnpkSEpwYm1keklHRmtaR2x1WnlCd1lYSjBhV04xYkdGeUlIVnVaR1Z5YzJOdmNtVnpEUW9nSUhOMFpYQnVkVzFpWlhJOU1Td2dJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ1VnZEdobElITjBaWEFnWW1WMGQyVmxiaUIwZDI4Z2JHbHVaUzF1ZFcxaVpYSnp
MaUJKWmlCcGRDZHpJREVzSUdWaFkyZ2diR2x1WlNCM2FXeHNJR0psSUc1MWJXSmxjbVZrRFFvZ0lITjBjbWx1WjNOMGVXeGxQVnhqYjJ4dmNudHpkSEpwYm1kc2FYUmxjbUZzWTI5c2IzSjlMQ0FnSUNBZ0pTQnpkSEpwYm1jZ2JHbDBaWEpoYkNCemRIbHNaUTBLSUNCMFlXSnphWHBsUFRJc0NTQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWxJSE5sZEhNZ1pHVm1ZWFZzZENCMFlXS
nphWHBsSUhSdklESWdjM0JoWTJWekRRb2dJSFJwZEd4bFBWeHNjM1J1WVcxbElDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDVWdjMmh2ZHlCMGFHVWdabWxzWlc1aGJXVWdiMllnWm1sc1pYTWdhVzVqYkhWa1pXUWdkMmwwYUNCY2JITjBhVzV3ZFhSc2FYTjBhVzVuT3lCaGJITnZJSFJ5ZVNCallYQjBhVzl1SUdsdWMzUmxZV1FnYjJZZ2RHbDBiR1VOQ24wTkNnMEtYR3h6ZEc1bGQy
VnVkbWx5YjI1dFpXNTBlMk52WkdWOVd6RmRXMTBsRFFvZ0lIdGNibTlwYm1SbGJuUmNiV2x1YVhCaFoyVjdYR3hwYm1WM2FXUjBhSDFjYldWa2MydHBjQ0FOQ2lBZ0lGeHNjM1J6WlhSN1ltRnphV056ZEhsc1pUMWNkSFJtWVcxcGJIbGNabTl2ZEc1dmRHVnphWHBsTEdaeVlXMWxQWE5wYm1kc1pTd2pNWDE5RFFvZ0lIdGNaVzVrYldsdWFYQmhaMlY5RFFvZ0lBMEtEUXBjYm1WM1kyO
XRiV0Z1WkNwN1hFTlBUVTFWVGtsVFRYMWJNVjE3SlEwS0lDQmNabTl1ZEhOd1pXTjdJekY5WEhONWJXSnZiSHNpTWpZeVJIMWNYQ1VOQ24wTkNnMEtEUXBjYm1WM1kyOXRiV0Z1Wkh0Y2NHRndaWEowYVhSc1pYMTdRMjl0YldsbFEyOXBiaUI3WEV4QlVrZEZJRnhqYjJ4dmNudHlaV1I5SUZ4RFQwMU5WVTVKVTAxN1JuSmxaVk5oYm5OOWZTQlRaV2w2YVc1bklIUm9aU0J0WldGd
WN5QnZaaUJqY25sd2RHOHRjSEp2WkhWamRHbHZibjBOQ2cwS1hHNWxkMk52YlcxaGJtUjdYR3hwYm10amIyeHZjbjE3YkdsbmFIUm5jbUY1ZlEwS1hHNWxkMk52YlcxaGJtUjdYSFJrZlZzeFhYdGNkRzlrYjF0cGJteHBibVZkZXlNeGZYME5DbHh1WlhkamIyMXRZVzVrZTF4aWNIMWJNbDE3WEc1dmFXNWtaVzUwSUNNeFhDQmNkR1Y0ZEdKbWV5TXlmVndnWENCOURRcGNibVYzWTI5dGJX
RnVaSHRjWW5Cd2ZWc3hYWHRjWW5CN0pGeHZjR3gxY3lSOWV5TXhmWDBOQ2x4dVpYZGpiMjF0WVc1a2UxeGljRzU5V3pGZGUxeGljSHNrWEc5dGFXNTFjeVI5ZXlNeGZYME5DbHh1WlhkamIyMXRZVzVrZTF4dFlXbHNmVnN4WFh0Y1pXMWhhV3g3WEdoeVpXWjdiV0ZwYkhSdk9pTXhmWHNqTVgxOWZRMEtEUXBjYVdad1pHWU5DaUFnWEdoNWNHVnljMlYwZFhCN0RRb2dJQ0FnSU
NCMWJtbGpiMlJsUFhSeWRXVXNJSEJrWm5ScGRHeGxQWHRjY0dGd1pYSjBhWFJzWlgwc0lIQmtabUYxZEdodmNqMTdUV0Z5YXlCU1lXRnpkbVZzWkhRc0lGUnBiU0JIZFdKdVpYSWdZVzVrSUVGaVpTQlhhWFJ6ZlN3TkNpQWdJQ0FnSUdOdmJHOXliR2x1YTNNOVptRnNjMlVzSUdKeVpXRnJiR2x1YTNNOWRISjFaU3dnYkdsdWEySnZjbVJsY21OdmJHOXlQVnhzYVc1clkyOXNiM0lzRFFvZ
0lDQWdJQ0JqYVhSbFltOXlaR1Z5WTI5c2IzSTlYR3hwYm10amIyeHZjaXdnZFhKc1ltOXlaR1Z5WTI5c2IzSTlYR3hwYm10amIyeHZjaXdOQ2lBZ2ZRMEtYR1pwRFFvTkNnMEtYR0psWjJsdWUyUnZZM1Z0Wlc1MGZRMEtEUXBjZEdsMGJHVjdYSEJoY0dWeWRHbDBiR1Y5RFFvTkNseHVkVzFpWlhKdlptRjFkR2h2Y25ON05uMGdEUXBjWVhWMGFHOXlldzBLWEdGc2FXZHVZWFYwYUc5eURRc
FFaWFJsY2lCMllXNGdTRzlzYkdGdVpGeGNEUW9nSUNBZ0lDQWdYR0ZtWm1Ga1pISjdRMWRKZlZ4Y0RRb2dJQ0FnSUNBZ1hHRm1abUZrWkhKN1FXMXpkR1Z5WkdGdGZWeGNEUW9nSUNBZ0lDQWdYR1Z0WVdsc2UyaHZiR0Z1WkdGQVkzZHBMbTVzZlEwS1hHRnNhV2R1WVhWMGFHOXlEUXBVYVcwZ1IzVmlibVZ5WkZ4Y0RRb2dJQ0FnSUNBZ1hHRm1abUZrWkhKN1ExZEpmVnhjRFF
vZ0lDQWdJQ0FnWEdGbVptRmtaSEo3UVcxemRHVnlaR0Z0ZlZ4Y0RRb2dJQ0FnSUNBZ1hHVnRZV2xzZTNScGJTNW5kV0p1WlhKQVkzZHBMbTVzZlEwS1hHRnNhV2R1WVhWMGFHOXlEUXBOWVhKNElIWmhiaUJTWVdGemRtVnNaSFJjWEEwS0lDQWdJQ0FnSUZ4aFptWmhaR1J5ZTBOWFNYMWNYQTBLSUNBZ0lDQWdJRnhoWm1aaFpHUnllMEZ0YzNSbGNtUmhiWDFjWEEwS0lDQWd
JQ0FnSUZ4bGJXRnBiSHR0TG5KaFlYTjJaV3hrZEVCamQya3VibXg5SUZ4aGJtUU5DbHhoYkdsbmJtRjFkR2h2Y2cwS1JHbGxaMjhnVUdGd1pYSnRZVzVjWEEwS0lDQWdJQ0FnSUZ4aFptWmhaR1J5ZTBOWFNYMWNYQTBLSUNBZ0lDQWdJRnhoWm1aaFpHUnllMEZ0YzNSbGNtUmhiWDFjWEEwS0lDQWdJQ0FnSUZ4bGJXRnBiSHQwYjIxbFFHTjNhUzV1YkgwTkNseGhiR2xuYm1GMWRHaHZ
jZzBLU21WbVppQkVaV0Z1WEZ3TkNpQWdJQ0FnSUNCY1lXWm1ZV1JrY250RWRXTnJSSFZqYTBkdmZWeGNEUW9nSUNBZ0lDQWdYR0ZtWm1Ga1pISjdRVzF6ZEdWeVpHRnRmVnhjRFFvZ0lDQWdJQ0FnWEdWdFlXbHNlMEIwYUdWeVpXRnNhbVZtWm1SbFlXNTlEUXA5RFFvTkNseHRZV3RsZEdsMGJHVU5DZzBLWEVGa1pGUnZVMmhwY0c5MWRGQnBZM1IxY21WQ1J5cDdKUTBLS
UNCY1FYUlFZV2RsVlhCd1pYSlNhV2RvZEh0Y2NtRnBjMlZpYjNoN0xWeG9aV2xuYUhSOWUxeG1jbUZ0Wlh0Y2FXNWpiSFZrWldkeVlYQm9hV056VzNkcFpIUm9QVFF1TldOdFhYdG9ZVzF0WlhKaGJtUnphV05yYkdVdWNHUm1mWDE5ZlgwTkNseEJaR1JVYjFOb2FYQnZkWFJRYVdOMGRYSmxRa2NxZXlVTkNpQWdYRUYwVUdGblpWVndjR1Z5VEdWbWRIdGNjbUZwYzJWaWIzaDdMVnh
vWldsbmFIUjllMXhtY21GdFpYdGNhVzVqYkhWa1pXZHlZWEJvYVdOelczZHBaSFJvUFRRdU5XTnRYWHRvWVcxdFpYSmhibVJ6YVdOcmJHVXVjR1JtZlgxOWZYME5DbHh0WVd0bGRHbDBiR1VOQ2cwS1hHSmxaMmx1ZTJGaWMzUnlZV04wZlEwS0RRcGNaVzVrZTJGaWMzUnlZV04wZlEwS0RRcGNhMlY1ZDI5eVpITjdmUTBLRFFwY2MyVmpkR2x2Ym50SmJuUnliMlIxWTNScGIyNTlYR3ho
WW1Wc2UzTmxZM1JwYjI0NmFXNTBjbTlrZFdOMGFXOXVmUTBLUTI5dGJXbGxJRU52YVc0Z2FYTWdZU0JpYkc5amEyTm9ZV2x1SUdOeWVYQjBieUJqZFhKeVpXNWplU0IzYVhSb0lHaDVZbkpwWkNCUWNtOXZaaUJ2WmlCVGRHVmhheUJoYm1RZ1VISnZiMllnYjJZZ1RHRmliM0l1SUZSb1pYSmxJR0Z5WlNCMGFISmxaU0IwZVhCbGN5QnZaaUIwYjJ0bGJuTTZJRU52YlcxdmJpd2dUV1ZrY
VhWdElHRnVaQ0JTWVhKbExpQlVhR1Z6WlNCaGJHd2dhR0YyWlNCbGNYVmhiQ0IyWVd4MVpTd2dZblYwSUhSb1pTQlNZWEpsSUhSdmEyVnVjeUJoY21VZ2JXOXlaU0JsY1hWaGJDQmhjeUIwYUdWNUlHRnlaU0IwYUdVZ2NtVmtaR1Z6ZENBb1lXNWtJR2hsYm1ObElHMXZjM1FnUTI5dGJYVnVhWE4wS1NCdlppQmhiR3dnZEc5clpXNXpMaUJEYUdWM2FXNW5JQ2h0YVc1cGJtY3BJRkpoY2
1VZ1kyOXBibk1nYVhNZ2JXOXlaU0IzYjNKcklIUm9ZVzRnWTJobGQybHVaeUJOWldScGRXMGdiM0lnUTI5dGJXOXVJR052YVc1ekxnMEtEUXBVYUdVZ1JtRnBjaUJEYjI1MGNtRmpkQ0J6ZVhOMFpXMGdZMkZ1SUdKbElIVnpaV1FnZEc4Z2FXMXdiR1Z0Wlc1MElHRndjR3hwWTJGMGFXOXVjeUJ2YmlCMGIzQWdiMllnZEdobElFTnZiVzExYm1semRDQkRhR0ZwYml3Z2FXNGdkMmhwWTJn
Z1lXeHNJR3hwYm10eklHRnlaU0JsY1hWaGJHeDVJSE4wY205dVp5NE5DZzBLUTI5dGJXbGxJRU52YVc0Z1oyRndjeUIwYUdVZ1luSnBaR2RsSUdKbGRIZGxaVzRnWTI5dGJYVnVhWE50SUdGdVpDQmpZWEJwZEdGc2FYTnRJR0o1SUhWemFXNW5JR0VnWm5WdVpHRnRaVzUwWVd4c2VTQmpZWEJwZEdGc2FYTjBhV01nWTNWeWNtVnVZM2tnZEc4Z2FXMXdiR1Z0Wlc1MElHTnZiVzExYm
1semRDQnBaR1ZoYkhNdUlBMEtEUXBjYzJWamRHbHZibnRKYlhCc1pXMWxiblJoZEdsdmJuME5DZzBLRFFwY2MyVmpkR2x2Ym50QmMzQmhjbUZuZFhOOVhHeGhZbVZzZTNObFkzUnBiMjQ2Wlhod1pYSnBiV1Z1ZEhOOURRb05DZzBLWEhOMVluTmxZM1JwYjI1N1UzbHpkR1Z0Y3lCVVpYTjBaV1I5RFFwT2IyNWxEUW9OQ2x4emRXSnpaV04wYVc5dWUxSmxjM1ZzZEhOOURRcFRiMjFsRF
FvTkNseHpaV04wYVc5dWUxVnVjbVZzWVhSbFpDQlhiM0pyZlEwS1ZHaGxJSEpwYzJsdVp5QndiM0IxYkdGeWFYUjVJRzltSUdOeWVYQjBieTFqZFhKeVpXNWphV1Z6SUhSb1pYSmxJR2x6SUdFZ2JHOTBJRzltSUhOamFXVnVkR2xtYVdNZ2QyOXlheUJrYjI1bElHOXVJR1JwWm1abGNtVnVkQ0JpYkc5amEyTm9ZV2x1SUdsdGNHeGxiV1Z1ZEdGMGFXOXVjeXdnY0hKdmIyWXRiMll0ZDI5eWF5
d2djMjFoY25RZ1kyOXVkSEpoWTNSeklHRnVaQ0J0WlhSaExXRnVZV3g1YzJsekxtZHBkbVZ6SUdFZ1ozSmxZWFFnWjNKdmRXNWtJR1p2Y2lCeVpXeGhkR1ZrSUhkdmNtc3VJRWh2ZDJWMlpYSXNJSFJvYVhNZ2MyVmpkR2x2YmlCamIyNTBZV2x1Y3lCMWJuSmxiR0YwWldRZ2QyOXlheTROQ2cwS1FXNHNJR1p2Y2lCcGJuTjBZVzVqWlN3Z1kyOXRjR3hsZEdWc2VTQjFibkpsYkdGMFpXUWdj
R0Z3WlhJZ2FYTWdkR2hwY3lCdFlYTjBaWElnY0dsbFkyVWdZbmtnU21WbVppQkVaV0Z1SUZ4amFYUmxlMlJsWVc0eU1EQTVaR1Z6YVdkdWMzMHVJRUoxZEN3Z2QyaGxjbVYyWlhJZ1NtVm1aaUJFWldGdUlHRndjR1ZoY25NZ1NtRnRaWE1nUkdWaGJpQmpZVzV1YjNRZ0RRcGNZbVZuYVc1N2FYUmxiV2w2WlgwTkNpQWdJQ0JjYVhSbGJTQkJJSEJoY0dWeUlHSjVJRXBsWm1ZZ1JHVmhia
UJjWTJsMFpYdGtaV0Z1TWpBd09XUmxjMmxuYm5OOURRb2dJQ0FnWEdsMFpXMGdRVzRnWVhKMGFXTnNaU0JoWW05MWRDQktZVzFsY3lCRVpXRnVJRnhqYVhSbGUycGhiV1Z6WkdWaGJuME5DbHhsYm1SN2FYUmxiV2w2WlgwTkNnMEtEUXBjYzJWamRHbHZibnREYjI1amJIVnphVzl1Y3lCY0ppQkdkWFIxY21VZ1YyOXlhMzBOQ2cwS1hITjFZbk5sWTNScGIyNTdVMlZzWmlCRmRtR
nNkV0YwYVc5dWZRMEtEUXBjYzNWaWMyVmpkR2x2Ym50R2RYUjFjbVVnVjI5eWEzME5DZzBLRFFwY2MyVmpkR2x2Ym50QmNIQmxibVJwZUgwTkNseHBibU5zZFdSbFozSmhjR2hwWTNOYmMyTmhiR1U5TUM0MVhYdGhjSEJsYm1ScFkybDBhWE5mY3pGZllYQndaVzVrYVhoZmFXeHNkWE4wY21GMGFXOXVMbXB3WjMwZ0tITnZkWEpqWlRvZ1hIVnliSHRvZEhSd2N6b3ZMM2QzZHk1d
mJtaGxZV3gwYUM1amIyMHZZMjl1ZEdWdWRDOHhMMkZ3Y0dWdVpHbGphWFJwYzE5aGNIQmxibVJsWTNSdmJYbDlLUTBLRFFwY1ltbGliR2x2WjNKaGNHaDVjM1I1YkdWN1lXSmljblo5RFFwY1ltbGliR2x2WjNKaGNHaDVlM1pzWkdKOUlDQU5DZzBLWEhOMVluTmxZM1JwYjI1N2IyWWdRMjl1WTJWd2RIME5DbE5sYm1RZ1lXNGdaVzFoYVd3Z2MzUmhkR2x1WnlBaVNTQmhiU0JoSUdO

dmJtTmxjSFFzSUhSb1pYSmxabTl5SUVrZ2RHaHBibXNpSUhSdklHTnNZV2x0SUhsdmRYSWdjbVYzWVhKa0xnMEtEUXBjWW1Gc1lXNWpaUTBLRFFwY1pXNWtlMlJ2WTNWdFpXNTBmUTBLDQoNClxiYWxhbmNlDQoNClxlbmR7ZG9jdW1lbnR9DQoNCg==

We have nothing to prove.

[34]{.underline}

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

With no time to lose, you head straight for the bridge to the Gates
Hillman Com plex. Before you get there, you feel a tug on your port
manipulator---a human has grappled onto you. "No human would so
ruthlessly efficiently neglect to explore every floor of Newell Simon
Hall before proceeding to the Gates Hillman Complex--- you must be a
robot!" The human, a Serious Businessperson, reprograms you to mine
the latest cryptocurrency, Numberwangcoin. You read the specification
of the Numberwangcoin protocol [8] and begin the exciting
hash-inverting guesswork.

switch (choose_dear_reader()) {

case 0: goto PAGE_51;

case 1: goto PAGE_51;

case 6: goto PAGE_51;

case 13: goto PAGE_51;

case 17: goto PAGE_51;

case e^π : goto PAGE_51;

case 42: goto PAGE_51;

case A_LOT: goto PAGE_68;

case ^{[−b±]{.underline}√}[b]{.underline}²[−4ac]{.underline}

_2a: goto PAGE_51;

case 1337: goto PAGE_51;

case 9001: goto PAGE_51;

case ONE_MILLION_DOLLARS: goto PAGE_51;

case 123456789101112: goto PAGE_51;

case ACKERMANN_5: goto PAGE_51;

case ℵ₁: goto PAGE_51;

}

35

That's Numberwangcoin!

Robert J. Simmons, Calculemus LLC, rob@calculem.us, Not From Somerset
7

Abstract

We present a new design for The Blockchain. This attempts to solve
several problems, including boredom, lost coins, shouty bits,
speculative investment, and the number 2 which you may remember from
school is deadly to humans.

Background: The Blockchain, Hashes, and Difficulty Every block in a
The Blockchain can be seen as the jamming together of three things:

1. The hash of the previous block

2. Some stuff you care about (The Ledger (TM)). In its simplest form,
The Ledger (TM) involves a bunch of addresses (big numbers) that
transfer value to one another; everyone can see The Ledger (TM) and
compute the current value of every address.

3. Some random stuff

Your job, as a miner in a The Blockchain, is to come up with random
stuff over and over, jam it together with the other bits and compute
its hash. A hash takes data and turns it unpredictably into a string
of, say, 256 bits. Then the hash is evaluated to see if it's Good™.

Being "Good™" is something that everybody working on the same The
Blockchain has to agree on: everybody has to be able to look at your
three parts, concatenate them themselves, compute the hash, and say,
"Yep, Julie's random stuff caused the jamming together to have a hash
that is Good™. Julie is a worthwhile member of society and deserving
of scarce resources."

Presumably Julie just found the Good™ hash by picking new random stuff
over and over until one of the versions of the random stuff was Good™.
Picking random stuff is like pulling the arm on a slot machine: it
produces some random output and that output might be Good™ news for
you.

A fundamental design aspect of any The Blockchain is difficulty. It
needs to become harder or easier to accidentally generate a Good™
block in order to keep the rate of newly solved blocks roughly
consistent across a The Blockchain. In Bitcoin's The Blockchain,
difficulty is recalibrated every 2016 blocks, with the goal of making
some contestant able to randomly come up with a jackpot Good™ random
value once every ten minutes.

In most The Blockchain, a Good™ hash has a lot of zeroes at the front.
This can lead to an important but subtle misconception that Good™ness
is a property of how many zero bits are at the start of the hash, and
difficulty is tweaked by calibrating how many zeroes there are at the
beginning of the hash.

Too easy: hash &
f800000000000000000000000000000000000000000000000000000000000000 is
zero Realistic: hash &
ffffffffffff0000000000000000000000000000000000000000000000000000 is
zero Unrealistic: hash &
fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff000 is
zero

In the "too easy" example, we would expect that it would only take 32
guesses at random stuff before one of our hashes would be Good™.

The only problem with saying that the difficulty is the number of
zeroes is that that the difficulty can then only get twice as hard or
twice as easy by adding or removing a bit-that-must-be-zero. The
better idea is to to say a Good™ hash is numerically smaller than some
target; lowering the target by a small amount increases difficulty a
small amount, in general.

1

36

Shouting Numberwang At Each Other

We turn to the problem we're solving. Specifically, the problem that
this isn't true enough:

So let's make it truer, and make Numberwangcoin in the process.
Computers in boring The Blockchain are actually shouting at each other
about programs with rather low hashes (hashes that are below the
target). Can we have them shout Numberwang?

Numberwang, Yes, But is it Numberwang Enough?

We could make the computer shouty bits a little more Numberwang by
requiring the inevitable zeroes in front of a hash to be the ASCII
representation of "NUMBERWANG NUMBERWANG NUMBERWANG" (all caps.
Remember: they're supposed to be shouting). This is a string with the
following 256-bit representation:

4e554d42455257414e47204e554d42455257414e47204e554d42455257414e47

The most uniform way to enforce this is to say that a computed hash
must be XORed against this shouty value before it is compared against
the target. Therefore, a Good™ hash becomes not the one that is
smallest in an absolute sense, but the one that is the Most
Numberwang. The hardest possible hash to come up with is no longer
zero, it is Numberwang (Numberwang Numberwang).

We encounter a problem, though. Even with a resources at a global
scale devoted to the problem of shouting boring Bitcoin low-value
hashes, the current target only has 18 leading zeroes.

In other words, if the Bitcoin protocol were based on our proposed
design, computers would regularly be shouting "NUMBERWAN" at each
other, but not necessarily "NUMBERWANG". At present, they would shout
a full "NUMBERWANG" about once every 32 transactions, though, so we
are close. But a truly climate-altering amount of computational
resources are being devoted to this shouty computer process. We need
to figure out how to make Numberwang with less.

Here we make the observation that we're using the ASCII encoding,
which falls in the range 0-127, wasting one bit per character. If we
truncate the first character, it becomes over a thousand times easier
to produce an actual full Numberwang, giving us an XOR value of

9d566c28b4abc19d1d04eab36145a55e0ce8e827559b0a2d2af06747413aacd8

This makes it 1024x easier to come up with Numberwang, and also allows
us to store an additional "num" and 4 bits of "b" in the shout string.
However, shouting a full "NUMBERWANG" in every new addition to The
Blockchain still corresponds to a hash difficulty¹that was only
reached in late 2016 on the Bitcoin network.

¹Roughly 270 trillion, for those following along at home. This means
that at the lowest difficulty setting, 1 in 270 trillion blocks would
have a full Numberwang.

2

37

It's evident that we need to go further. We will compress even further
using the predictable "\"A\" is zero, \"B\" is one. . . " encoding, in
which we need five bytes, instead of seven, per letter.

Char: N U M B E R W A N G N U M B E R W A N Ord: 13 20 12 1 4 17 22 0
13 6 13 20 12 1 4 17 22 0 13 Binary:
0110110100011000000100100100011011000000011010011001101101000110000001001001000110110000000110100Hex:
6 d 1 8 1 2 4 6 c 0 6 9 9 b 4 6 0 4 9 1 b 0 1 a

This encoding allows us to encode "NUMBERWANG" in 50 bits. Given that
we have 256 bits to work with, and given that the last six bits
represent a truly astronomical difficulty, we will leave the last six
bits zero, meaning that our shouty XOR-with-the-hash value that is:

6d181246c0699b460491b01a66d181246c0699b460491b01a66d181246c06980

I Can't Think Of Any More Numbers

Getting a single numberwang in this encoding represents a
difficulty²reached in early 2011, way before anyone gave a crap
about any of this. Still a bit high, though: at the lowest difficulty
setting, only 1 in every 260 thousand blocks would be expected to
shout a full "Numberwang."

We choose to live with this reality, and turn "lemons" into Wordwang.
Observe that the difficulty recalibration interval represents a
natural time demarcation (corresponding to roughly two weeks in
Bitcoin's The Blockchain). If, between two recalibration steps, no
full 50-bit numberwang appears, we enter Sudden Death.

In Sudden Death, we re-hash the hash of the previous board
recalibration block to get the Deadly Number Gas Hash (Deadly Gash).
The address with a non-zero balance that matches the most bits of the
Deadly Gash, starting with the least-significant digit, will have its
balance replaced by 2 WangerNumb. If there's a tie, all first place
winners lose.

The Sudden Death protocol will make adoption of Numberwangcoin faster,
because the more people mine Numberwangcoin, the faster the difficulty
will rise to a level that makes Sudden Death at first unlikely, and
then, in time, impossible.

Let's Rotate The Board!

The most important and lasting effect that happens along with
difficulty recalibration is Rotating the Board. When we Rotate the
Board, every address with a nonzero balance is put in numerical order
by address (not balance). We then rotate value from an address to its
next highest address. The highest address wraps around and transfers
to the lowest.

If an account has more than a thousand WangerNumb, then the balance is
rounded down to the next power of 10, and one-one-thousandth of that
amount is rotated to the next address on the board. If an account has
less than a thousand WangerNumb, then one full WangerNumb is rotated
to the next address on the board until the balance becomes zero and it
leaves the rotation.

This wealth redistribution mechanism ensures that no Numberwangcoin
value will ever truly leave circu lation. It also makes Numberwangcoin
a terrible mechanism for long-term investment, charging approx imately
2.5 percent in redistributive taxation every year. These fees are
quite trivial if one is holding Numberwangcoin for a short period of
time as a medium of free exchange, bringing The Blockchain back to the
purpose that I, um I mean Satoshi, intended.

The Maths Coin That Simply Everyone Is Talking About

The genesis block for Numberwangcoin, along with a state-of-the-art
browser-based miner, will or will not be available from
https://numberwangco.in/ on March 30, 2018.

²About 260k for those following along at home.

3

38

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

You search the room in vain for thread-like objects. Eventually, you
realize that the thread-like objects were inside you all along: wires!
While unscrewing your main casing, you load and analyze your wiring
diagram. You look down upon your own wires, identify one that your
analysis determined is not critical to your functionality, carefully
place it into the jaws of the scissors from your sewing kit, and cu---

Segmentation fault (core dumped)

39

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

"Don't worry, you've got this!"

{width="4.277777777777778in"
height="1.3472222222222223in"}

switch (choose_dear_reader()) {

case FANTASTIC:

Press the up, down, left, and right pads in that order.

goto PAGE_69;

case WAY_OFF:

Press the left, right, left, and right pads in that order.

goto PAGE_178;

case MISS:

Don't press any pads.

goto PAGE_28;

}

40

Stochastic Processes

Portrait of Markov

8 Ritwik density estimation and analysis using real techniques Ritwik
Gupta, Ritwik Das, and Ritwik Rajendra

Keywords: Ritwik, Council of Ritwiks, population den sity, statistical
analysis, deep learning, state of

the art, monte carlo, bayesian methods, real

computational methods and statistics

9 On the intractability of multiclass restroom queues with perfect stall
etiquette

Sarah Allen and Ziv Scully

Keywords: Markov chain, recursive renewal-reward, poop [41]{.underline}

8

Ritwik Density Estimation and Analysis Using Real Techniques

{Ritwik, Ritwik, Ritwik} Gupta, Das, Rajendra^∗

{ritwikg1, rsdas, ritwikr} @ andrew.cmu.edu

The Council of Ritwiks @ Carnegie Mellon

Abstract

The distribution of Ritwiks across the world is a question pursued by
countless researchers across a variety of fields. A yet unanswered
question¹, we seek to once and for all put this question to rest. We
also provide auxiliary discussion and proofs demonstrating various
statistical properties of the Ritwik population.

1 Distribution of Ritwiks

Comprehensive, boots on the ground research was done to effectively
determine the distribution of Ritwiks across the world. Using
Facebook², we were able to ascertain the location of and establish
contact with Ritwiks everywhere (see Figure 1). We collected a large
sample size (N = 12) and used Hamiltonian Monte

{width="3.0014173228346457in"
height="1.85087489063867in"}

Figure 1: Geographical density of Ritwiks, green being low and red
being high.The white areas denote the authors were lazy to make a heat
map covering the globe.

Carlo methods to simulate certain parameters that are backed by
Bayesian goodness, undeniably proving that our method is rock solid.
We cast a net out to collect the samples , the deep kind of net
therefore guaranteeing the best sample. All data was analyzed with
cutting edge tools [1, 2]. The following math not only looks cool,
but makes reviewers think that we did real work because math makes it
look that way.

Count_i ∼ P oi(λ), (1a)

λ ∼ DiscreteUniform(0, 7e9), (1b)

Z

_π(q)f(q)^X V ar_π[f]

_ESS (1c)

λ

^∗Authors are listed in the order of narcissism towards their first
names

¹https://scholar.google.com/scholar?hl=en&as
[s]{.underline}dt=0%2C39&q=distribution+of+ritwiks&btnG=
²https://facebook.com

[42]{.underline}

1.1 Estimating the Density of "Ritwik" Using Novel Methods

Ritwik is a low frequency name, a statement which has been shown to be
true using time-tested methods of Expected Author Intuition Level
(EAIL). Li et. al. [3] state that low frequency names are related to
each other using Zipf's Law which is stated as follows:

Let X = {x₁, x₂, x₃, x₄, x₅, x₆, x₇, x₈, x₉, x₁₀,
x₁₁, ..., x_N }, N = some large number and X is the vector of
names present in the world. Let Y = {y₁, y₂, y₃, ...y_N } be
the ranking of each x_i. Therefore, Zipf's Law states that:

_{Y ∝}1

_Count(X)+ ξ (2)

We completely ignore this rule and use deep learning since
representation learning solves all problems. Assume a low density
uniform prior on the density of Ritwik over geographical locations
(which are sorted alphabetically and mapped to whole numbers). When
you imagine it in your head, it sure does look like a line, right?
Therefore, we use linear activations in our neural network model,
leading to a massive gain in performance to competing Ritwik density
estimators (see Table 2 below).

---

Method MAP MRP GDP

---

SVM 0.05 0.02 9.65

SVM + BBN 0.45 0.21 3.21

RBM 0.68 0.44 2.96

Linear NN (ours) 1.00 0.99 0.01

---

Table 1: Performance of Ritwik density methods.To generate this table
we used a rigorous foolproof exper imentation technique called YOLO
(You Only Lie Once).

An example of an architecture we did not use is included below as
reference, carefully created in MS Paint for the highest quality
rendering and production value.

{width="4.001917104111986in"
height="2.4749442257217846in"}

Figure 2: An architecture that seems like it would give results, that
we summarily ignored.

To make our results reproducible, we have stuck to well-backed
academic practices of releasing all of our code on private GitHub
repositories only accessible via an email to one of our auxiliary
email addresses that we check once in a blue moon, or after we publish
everything of use from the dataset.

[43]{.underline}

2 Popularity of "Ritwik" Over Time

Though Ritwiks themselves are insanely popular³, the name Ritwik
itself has not seen widespread gain in usage throughout history. Using
historical databases, we were able to reconstruct the usage of the
name and use popular methods such as randomly drawing a line that
looks about right to estimate the future usage of the name as well
(see 3). As evident, the name Ritwik is predicted to skyrocket as this
paper is made public. Eventually, all people will be named Ritwik, and
the universe will be at peace.

Figure 3: The occurrence of the name Ritwik over time. Green line
represents the year this paper was published.

3 On the Immortality of Ritwiks

Based on the vast quantity of Ritwiks we have met, none of them have
been dead or deceased. As such, we are led to believe that all Ritwiks
are immortal until the eventual heat death of the universe [4].

Lemma 1. Given any Ritwik, the average lifespan of the individual will
be ∞.

Figure 4: Search of the U.S. Social Security Death Index for "Ritwik".

³Refer to our peers.

[44]{.underline}

Proof. Let us assume that all Ritwiks die, for the sake of
contradiction. Therefore, a record of death must exist within the
United States Social Security Death Index⁴. However, we can see in
Figure 4, no records of

deceased Ritwiks exist. Therefore, Lemma 1 must hold.

---

---

4 Adversarial Ritwiks

With the recent successes in people being able to finally spell our
name properly, adversarial attacks against our nomenclature have
become prevalent. Simple affine transformations often result in
massive confusion amongst peers and colleagues. An example of these
transformations can be seen in Table 2 below. Many

---

Transformation

---

Ritvik

Ritwick

Rick

Hrithik

"How about I call you Rob?"

---

Table 2: Example of affine transformations on the name "Ritwik"

defenses exist against adversarial attacks against the name "Ritwik".
Papernot et. al. [5] suggests that distilling these toxic people out
of your life demonstrates a sizable increase in the quality of life.
However, many attacks have been shown directly bypassing distillation,
which means that you're stuck with hearing various people call you
different names for the rest of your life, which as shown in Section
3, is forever.

5 Conclusion

We have demonstrated absolutely nothing of use, but are still proud of
our contribution to the world. If you are a Ritwik and are currently
not a member of the Council, please email us at once to rectify this
grave mistake. If you are currently not named Ritwik and would like to
be a member of the Council, please refer to your country's name change
applications. No Ritwiks were harmed in the making of this paper.

References

[1] Matei Zaharia, Mosharaf Chowdhury, Tathagata Das, Ankur Dave,
Justin Ma, Murphy McCauley, Michael J. Franklin, Scott Shenker, and
Ion Stoica. Resilient distributed datasets: A fault-tolerant
abstraction for in-memory cluster computing. In Proceedings of the 9^th
USENIX Conference on Net worked Systems Design and Implementation,
NSDI'12, pages 2--2, Berkeley, CA, USA, 2012. USENIX Association.

[2] Fonnesbeck C. Salvatier J., Wiecki T.V. Probabilistic
programming in python using pymc3. In PeerJ Computer Science, 2016.

[3] Wentian Li. Analyses of baby name popularity distribution in
u.s. for the last 131 years. 18:1, 09 2012. [4] Chas A. Egan and
Charles H. Lineweaver. A larger estimate of the entropy of the
universe. 2009.

[5] N. Papernot, P. McDaniel, X. Wu, S. Jha, and A. Swami.
Distillation as a defense to adversarial perturbations against deep
neural networks. In 2016 IEEE Symposium on Security and Privacy (SP),
2016.

⁴http://search.ancestry.com/search/db.aspx?dbid=3693

[45]{.underline}

CONFIDENTIAL COMMITTEE MATERIALS
{width="1.000485564304462in"
height="1.000485564304462in"}

SIGBOVIK 2018 Paper Review

Paper 5: Ritwik Density Estimation

and Analysis Using Real Techniques

Richard Robertson

Rating: A Fine Paper

Confidence: Germanic

As someone who gets bugged by an endless sequence of "can I just call
you Ritwik Ritwik" requests, I find this paper to be Relatable
Content^TM.

46

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

Natural language processing is a difficult problem, but as a
state-of-the-art fic tional robot, geometric idioms with non-geometric
meanings, such as "take it back" meaning "apply the inverse of the
most recently applied operator", are robot-with factory-settings's
play. You therefore roll your robot wheels rightwards in perfect
execution of the expected dance mo---

Crash!

You bump into a backwards-moving human, who yells in surprise and
spills her drink on you. It appears you did not perform the expected
dance move. Worse still, the drink is dangerously conductive! You must
locate a cleaning station and remove the spill before you
short-circuit. Fortunately, the human is willing to help: she
spontaneously apologizes for not seeing you there and offers to show
you to a "bathroom" to clean up. Hoping that a bathroom contains
suitable cleaning supplies, you follow her.

Upon entering the human bathroom, you discover it to be a curious
colocation of cleaning stations and waste disposal stalls. You wipe
off the dangerously conductive beverage and decide that, while you are
here, you may as well empty Waste Disposal Bays #1 and #2. After
reviewing a recent paper on human waste-disposal protocol [1], you
enter a stall. As you position your Waste Disposal Bays, you hear the
door open: another human is here for waste disposal.

switch (choose_dear_reader()) {

case WAIT_FOR_IT:

Empty Waste Disposal Bay #1.

goto PAGE_153;

case GO_FOR_IT:

Empty Waste Disposal Bays #1 and #2.

goto PAGE_206;

}

[47]{.underline}

On the Intractability of Multiclass Restroom Queues

with Perfect Stall Etiquette

9

Sarah Allen

Large Internet Company

Nabisco Factory, Bakery Square

Pittsburgh, PA

ABSTRACT

Ziv Scully

Large Computer Science Department Yet Another Gates Building, Schenley
Park Pittsburgh, PA

Algorithm 2.1 Protocol for Class 2 Customers

We extend prior work on queueing-theoretic bathroom humor. Our results
aren't as good, but the system model is funnier.

ACH Reference format:

Sarah Allen and Ziv Scully. 2018. On the Intractability of Multiclass
Re stroom Queues with Perfect Stall Etiquette. In Proceedings of
SIGBOVIK 2018, Pittsburgh, PA, USA, March 29, 2018 (SIGBOVIK '18), 2
pages.

1 INTRODUCTION

The complex social dynamics of homo sapiens results in intricate
etiquette protocols for many activities, including restroom usage.
Such protocols are a significant mathematical obstacle for queueing
theorists who wish to rigorously analyze the performance metrics of
restrooms. A recent breakthrough by Gardner and Scully [2] presented
the first theoretical analysis of a restroom queue that accounted for
restroom etiquette. The work addresses the so-called M/M/3/C2UPN,
which handles the case of urinals in a men's re stroom under the usual
rule: no two adjacent urinals may be simul taneously occupied.

By now, the careful reader will have noticed that Gardner and Scully
[2] consider only one of the two traditional genders served by
multioccupancy restrooms and only one of its two customer classes
[3]. Men, as notoriously simple creatures, employ a urinal protocol
that admits exact analysis in almost alarming generality. In contrast,
in this work we show that the stall protocol employed in women's
restrooms results in a Markov chain whose behavior is impossible to
exactly analyze using known techniques, except under very specific
conditions. The intractability arises from the convoluted interaction
between customers of both classes, which has not been considered in
prior work.

2 SYSTEM MODEL

We consider a women's restroom with k servers, namely stalls. Customers
arrive with a Poisson process of rate λ. Each customer is independently
Class 1, with probability p₁, or Class 2, with prob ability p₂ = 1 −
p₁. Class 1 customers have an exponential service time distribution of
rate µ₁, and similarly for Class 2 with rate µ₂. However, the story
for Class 2 customers, described in detail below, is complicated due to
the following restriction.

A Class 2 customer can only be served while it is

the only customer of any class in the system.

That is, in a women's restroom, no one can hear you poop. This
protocol ensures that every customer can plausibly maintain the facade
that they are not a Class 2 client [4].

The protocol for Class 2 customers is defined formally in Algo rithm
2.1. We now describe the intuition behind the protocol. A

Begin as Inactive.

• If there is a Class 1 customer at another server, stall, namely do
nothing.

• If only Class 2 customers occupy other servers, begin a sitoff,
abandoning at rate ν₂.

ś If Number-2-ing, instead do not abandon.

ś If one of the other customers is Number-2-ing, instead abandon at
increased rate ν₂ + ξ₂.

• If there are no other customers in the system, permanently
transition from Inactive to Number-2-ing.

ś While Number-2-ing with no other customers in the system, complete
service at rate µ₂.

Class 2 customer, instead of beginning service immediately upon
entering a server, initially stalls, or blocks, until all k servers
are empty or contain other Class 2 customers. We conjecture this be
havior is the namesake of the colloquial term for servers. Once only
Class 2 customers remain at servers, they begin a sitoff, dur ing
which each customer may leave, dethroning themselves as a contender to
be the lone Class 2 customer to receive service. This occurs at
stochastic rate ν₂, and the leaving customer has to find a new place
to number-2. Until the queue is empty, the Class 2 cus tomers at the
servers alternate between stalls and sitoffs, depending on whether
there is a Class 1 customer in service.

If the system is stable, eventually a single Class 2 customer will
occupy the system, at which point they finally begin service. They
become the sole number-2-ing customer. As other customers arrive, they
occupy servers as normal, with other Class 2 customers expe riencing
stalls and sitoffs as normal. The number-2-ing customer stalls during
sitoffs between the other Class 2 customers. A suspi cious air about
the number-2-ing customer gives sitoff participants a chance to sniff
them out. When a sitoff participant discovers a number-2-ing customer,
they know they will not win the sitoff, so they abandon the system.
This discovery happens at rate ξ₂, so the abandonment rate of sitoff
participants in the presence of a number-2-er to ν₂ + ξ₂.

3 INTRACTABILITY OF EXACT ANALYSIS We can describe the system as a
Markov chain whose states are 4-tuples of natural numbers
(n,s₁,s₂,д):

• n is the number of customers in the queue,

• s₁ is the number of Class 1 customers at a server, • s₂ is the
number of Class 2 customers at a server, and • д₂ is the number of
Class 2 customers number-2-ing.

[48]{.underline}

SIGBOVIK '18, March 29, 2018, Pi sburgh, PA, USA Sarah Allen and Ziv
Scully

The states are divided into those in the repeating portion, which have n
≥ 1, and those in the initial portion, which have n = 0. States in the
repeating portion obey the constraint s₁ + s₂ + д₂ = k, and those
in the initial portion obey s₁ + s₂ + д₂ ≤ k. All states obey д₂
≤ min{s₂, 1}.

Transitions out of states in the repeating portion of the Markov chain
are as follows:

• (n,s₁,s₂,д) → (n + 1,s₁,s₂,д) at rate λ, due to an arrival;
• (n,s₁,s₂,д) → (n−1,s₁,s₂,д) at rate p₁s₁µ₁, due to a
Class 1 completion and the next customer being Class 1;

• (n,s₁,s₂,д) → (n − 1,s₁ − 1,s₂ + 1,д) at rate p₂s₁µ₁,
due to a Class 1 completion and the next customer being Class 2; • (n,
0,s₂,д) → (n − 1, 1,s₂ − 1,д) at rate p₁s₂(ν₂ + дξ₂), due
to a Class 2 abandonment and the next customer being Class 1; and

• (n, 0,s₂,д) → (n − 1, 0,s₂,д) at rate p₂s₂(ν₂ +дξ₂), due
to a Class 2 abandonment and the next customer being Class 2.
Transitions out of states in the initial portion are routine to state
and thus omitted. The highlight is the transition (0, 0, 0, 1) → (0, 0,
0, 0)

at rate µ₂, due to a Class 2 customer's completion. The analysis of
Gardner and Scully [2] took advantage of the recursive
renewal-reward (RRR) technique [1]. The RRR technique applies,
roughly speaking, when the states in the repeating por tion can be
partitioned into layers such that transitions between layers form a
directed acyclic graph. Our system's Markov chain has one layer for
each triple (s₁,s₂,д). The layers form two con nected components,
one for д = 0 and another for д = 1. Unfor tunately, both components
have cyclic transitions between layers: (1, k − 1, 0) ↔ (0, k, 0) and
(1, k − 2, 1) ↔ (0, k − 1, 1). We have seen that RRR cannot exactly
solve this Markov chain. Similar issues occur when attempting matrix
analytic methods and other techniques. A glimmer of hope comes from
Gardner and Scully [2], who found that RRR applies to a 5-urinal
system, which also had cyclic transitions between layers of its Markov
chain. This is because for each transition from layer A to layer B in
that Markov chain, the total rate of transitions towards the initial
portion never increases going from layer A to layer B, ensuring the
existence of some matrix's square root or something¹. This yields
the following conclusion.

Theorem 3.1. We can only analyze the present system if 3ν₂ =
2(ν₂ + ξ₂) = µ₁.

And that is just a ridiculously specific assumption, even for a
SIGBOVIK paper.

4 SUGGESTED PROTOCOLS

Here we present some alternative strategies for rendering the above
system analyzable and suggest that customers of women's restrooms
implement them so that we can rigorously demonstrate the suboptimality
of the system.

Get Your Shit Together. Class 2 customers stall while Class 1 cus
tomers remain in the system. When a potential standoff is reached, all
Class 2 customers simultaneously initiate service. Note: this tech
nique has been observed in practice, as long as no two customers

¹Lossy personal communication from past Ziv to present Ziv.

occupy the common area at the same time, thus assuring plausible
anonymity (unless, of course, the bathroom is in a computer science
department, where number of clients who use women's restrooms is
regrettably low).

Shit or Get Off the Pot. If the customer encounters a situation in
which they would like to run a Class 2 job, but cannot doo-doo due to
other customers in the system, they must immediately call process Not
Giving a Shit or process Full of Shit, both of which are defined
below.

Not Giving a Shit. The customer decides that their re-poo-tation is
worth tarnishing for the purposes of optimality and brazenly uses the
available resource, regardless of the state of other servers.

Full of Shit. The customer refrains from using the Class 2 services
provided by the public restroom and blocks until they can use a
guaranteed private resource at home².

Shit Yourself. Not recommended.

REFERENCES

[1] Anshul Gandhi, Sherwin Doroudi, Mor Harchol-Balter, and Alan
Scheller-Wolf. 2013. Exact Analysis of the M/M/K/Setup Class of Markov
Chains via Recursive Renewal Reward. In Proceedings of the ACM
SIGMETRICS/International Conference on Measurement and Modeling of
Computer Systems (SIGMETRICS '13). ACM, New York, NY, USA, 153--166.

[2] Kristen Gardner and Ziv Scully. 2017. RRR for UUU: Exact
Analysis of Pee Queue Systems with Perfect Urinal Etiquette. In
Proceedings of SIGBOVIK 2017, Pittsburgh, PA, USA, March 31, 2017
(SIGBOVIK '17). ACH, 163--167.

[3] Taro Gomi and Amanda Mayer Stinchecum. 1993. ¯ Everyone Poops.
Kane/Miller Book Publishers.

[4] Scout Ysabella Reid. 2013. It's True, Girls Don't Poop! (2013).
https://www. theodysseyonline.com/true-girls-dont-poop

²In the first named author's experience, this approach yields poor
results when one lives in a dormitory hall with a common restroom.

[49]{.underline}

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

"Not bad for your first time!"

{width="4.277777777777778in"
height="1.3472222222222223in"}

switch (choose_dear_reader()) {

case FANTASTIC:

Press the left, right, down, and right pads in that order.

goto PAGE_87;

case EXCELLENT:

Press the right, left, right, and down pads in that order.

goto PAGE_69;

case MISS:

Don't press any pads.

goto PAGE_28;

}

50

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

Hmmmm, not quite. Guess again!

switch (choose_dear_reader()) {

case 0: goto PAGE_51;

case 1: goto PAGE_51;

case 6: goto PAGE_51;

case 13: goto PAGE_51;

case 17: goto PAGE_51;

case e^π : goto PAGE_51;

case 42: goto PAGE_51;

case A_LOT: goto PAGE_68;

case ^{[−b±]{.underline}√}[b]{.underline}²[−4ac]{.underline}

_2a: goto PAGE_51;

case 1337: goto PAGE_51;

case 9001: goto PAGE_51;

case ONE_MILLION_DOLLARS: goto PAGE_51;

case 123456789101112: goto PAGE_51;

case ACKERMANN_5: goto PAGE_51;

case ℵ₁: goto PAGE_51;

}

51

52

Ayyy Eye

A erimage of a Crimson Eye

10 PSYCHO: PerSonalitY CHaracterizatiOn of artificial intel ligence

Achal Dave and Rohit Girdhar

Keywords: interpretability, psychology, deep learning, artificial
intelligence, rorschach

11 The NUGGET non-linear piecewise activation Stephen Merity

Keywords: deep learning, neural networks, nugget, nuggets, chicken
nuggets, smart

12 Substitute teacher networks: Learning with almost no su pervision

Samuel Albanie, James Thewlis, and João F. Henriques Keywords:
substitute, teacher, networks

[53]{.underline}

PSYCHO: PerSonalitY CHaracterizatiOn of artificial intelligence 10

Achal Dave

Cranberry-Lemon University

Abstract

Recent times have seen great advancements in the field of AI, thanks to
the resurgence of deep learning. It has impacted virtually every aspect
of our lives, from generat ing new cat videos [4], to converting cat
videos into dog videos [2]. However, these advancements have also
stoked fear in the hearts of us humans: what if the robot hand that
learned to open door knobs instead decides to use its skills to pick up
a gun and point it at us? Needless to say, the solu tion is not fewer
guns, but the mental health of these robots. In this work, we try to
assuage those concerns by proposing a method to analyze the brains of
our robots. Our method takes years of human psychology research and
brainlessly applies it to analyze the deep networks that form the funda
mental cognitive system of modern day robots. We evaluate our method on
the latest and greatest deep networks and uncover the ones most likely
to 'break bad'.

1. Introduction

"AI is a fundamental risk to the existence of human civilization."

Elon Musk (July 2017)

"I was trying to turn off some lights and they kept

turning back on. After the third request, Alexa stopped responding and
instead did an evil laugh."

Reddit user (January 2018)

"The #BostonDynamics #robots are learning. Soon they'll be opening our
fridges and stealing our beer."

Dr. Randy Olson (February 2018, via Twitter)

Lets face it. The threat of AI is real, and the leaders of our tech
industry have gone out of their way to warn us about it. However, the
lack of tools to interpret our AI meth ods has tied the hands of AI
researchers, forcing them to fo cus on making their methods stronger
with no regard to the

Rohit Girdhar

Cranberry-Lemon University

{width="0.5521533245844269in"
height="0.5737215660542432in"}

{width="0.6556824146981627in"
height="0.5694072615923009in"}{width="2.2431255468066493in"
height="1.3458727034120734in"}Figure 1. When will AI go haywire?
Understanding how AI will act in the future requires a carefully
designed psychological anal ysis using the widely acclaimed Rorschach
ink blot test.

future of humanity. This problem is especially dire in the field of deep
learning, where the dark magic of stochastic gradient descent carves out
ultra high dimensional spaces to learn representations unimaginable by
humans. In this work, we take a step back and attempt to analyze the
think ing process of the deep networks we have crafted, before it is too
late.

Today, the Turing test is largely solved [1, 5, 3]. Our method, PSYCHO
instead uses the Rorschach inkblot test to analyze artificial
intelligence. The test works by show ing an inkblot image, like in Table
1 (column 1), and asks the user to pick a sentence that best describes
that inkblot from 7 options (we follow the paradigm from http://
theinkblot.com/). We design an approach to allow state of the art deep
networks to take this test, by finding nearest neighbors of their
representation with a representa tion for each option. We report some
insightful analysis of these networks in Sec. 3.

2. Approach

The Rorschach ink blot test, as presented on http: //theinkblot.com/,
requires the test-taker to pick a sentence describing each of the 10
Rorschach ink blots. Un fortunately, despite our best efforts, we were
unable to coax current AI models into taking online personality tests.

Undeterred, we developed a novel approach for psycho [54]{.underline}

logically evaluating our models. For each ink blot, we col lected an
image representing each potential response (such as "a giraffe in a
bathtub"). Unfortunately, naively collect ing images can lead to a bias
in the selected images. To overcome any such bias, we directly query
Google Image Search for an unbiased list of images for each potential re
sponse. We then selected a single image from these results for each
response query while trying our very hardest not to

Model Sickness Notes

---

47%

---

75%

78%

60%

---

A-net "Positive attitude towards ev erything"

"very annoying"

V-net "aspire to [be] CEO", "horrible bore"

I-net "short attention-span", "work very slowly"

use our personal biases.

Armed with this dataset, we present each ink blot along with potential
responses to our model, and select as a re sponse the image that the
model thinks is most like the ink

R-net D-net

"succeeded beyond wildest dreams",

"frequently mentions paradigm shifts"

blot.¹

3. Experiments

We present qualitative and quantitative results, along with
psychological notes for five popular Convolutional Neural Network models
in the computer vision community. We have anonymized the names to
protect against lawsuits avoid upsetting anyone.

In Table 1, we present the extensive analysis provided by
http://theinkblot.com. We immediately notice that our models have
surprisingly varied personalities. "A net" is a prototypical optimist,
or what experts may refer to as "the SpongeBob". V-net and I-net share a
high sick ness quotient, which we explore further through qualitative
results.

Unfortunately, trusting experts can mislead our under standing of
potential societal threats. To overcome this, we present the raw results
from our method in Table 2 for fur ther public analysis.

Disturbing responses: While some responses from our model are playful
(e.g. Table 2 Row 5), there are numer ous worrying signs in their
responses. I-net, in particular, consistently chooses disturbing imagery
(a satanical head in Row 3, a satanical eye in Row 5, a strange creature
in Row 7, and what is indubitably a satanical ritual in Row 9). Equally
worrying is the creepy imagery provided as re sponses by V-net, R-net,
and D-net in Row 1 (a monsterous face) and, worse, in Row 5 (a
Teletubby).

Intellectual diversity: The lack of diversity in AI is plainly visible
from our analysis. In particular, we discover for the first time that
models developed in the same insti tution (R-net and D-net) develop
equivalent psychological tendencies.

4. Conclusion

While we are far from preventing the inevitable AI apoc alypse, we
believe our method will go a long way in en

¹In particular, we take the final layer representation of the ink blot
and all response images, and choose the response that minimizes
Euclidean distance to the ink blot. We hope to publicly release our
code.

Table 1. Quantitative and qualitative results from the Rorscahch test,
according to one online test.

abling AI researchers to psycho-analyze their deep net works before
deploying them to read every single Snapchat we post through the day.

N.B.: This paper is a work of satire and should not be taken seriously.

References

[1] Computer ai passes turing test in 'world first'. http://
www.bbc.com/news/technology-27762088, 2014.

[2] J.-Y. Z. et al. CycleGAN. https://github.com/ junyanz/CycleGAN,
2017.

[3] L. Hardesty. Computer system passes "visual turing test".

[4] J. Johnson. Meow generator: This deep learning AI generated
thousands of creepy cat pictures. Motherboard, 2017.

[5] C. Osborne. Mit's artificial intelligence passes key turing test.
http://www.zdnet.com/article/

mits-artificial-intelligence-passes-key-turing-test/, 2016.

[55]{.underline}

Query A-net V-net I-net R-net D-net

Query A-net V-net I-net R-net D-net Table 2. Qualitative results on
the Rorschach test.

[56]{.underline}

The NUGGET Non-Linear Piecewise Activation [11]{.underline}

Stephen Merity ¹

Abstract

The choice of activation functions in deep neural networks has a
significant impact on the train ing dynamics, task performance, and
potential acronyms of resulting work. While numerous ac tivation
functions have been proposed, such as the Rectified Linear Unit
(ReLU), most are de rived from the domain of mathematics rather than
by drawing inspiration from nature. We pro pose a non-linear piecewise
activation function, the NUGGET activation function, which is a re
sult of a complex zero-sum pricing game refined over decades of
multi-agent interaction simula tion. We verify the effectiveness of
the activa tion by experimental analysis on the Modified National
Institute of Standards and Technology (MNIST) digits task (Neural
Numerology) and achieve state of the art results¹.

1. Introduction

The need for effective activation functions has fueled a rapid
exploration of all mathematical functions. This is problematic for those
of us still scared of mathematics. As such, a counter culture of human
curated artisanal activa tion functions has emerged.

Dropout (Srivastava et al., 2014) may be the first instance of a human
curated artisanal regularization technique that entered wide scale use
in machine learning. Dropout, sim ply described, is the concept that if
you can learn how to do a task repeatedly whilst drunk, you should be
able to do the task even better when sober. This insight has resulted in
nu merous state of the art results and a nascent field dedicated to
preventing dropout from being used on neural networks.

Our work seeks inspiration from the natural world in pro viding new and
intuitive manners to frame and explore re cent neural network advances.
In the following sections we analyze a specific subset of these
naturally occurring activation and regularization techniques, which we
shall broadly refer to as NUGGET functions, to understand the impact
they may have when applied to neural networks.

¹Our state of the art results can be seen as state of the art results
by ignoring the current state of the art.

2. The NUGGET n-player zero-sum game

The chicken nugget was invented in the 1950s by Robert C. Baker, a
food science professor at Cornell University, and published as
unpatented academic work. Since then, it has been a pivotal component
in the raging fast food wars that have beseiged the nations across
earth. Speculation exists that SpaceX (Musk, 2002) was started in an
attempt to escape the ever looming threat of NUGGET warfare. Given the
intense research, both theoretical and experimen tal, in determining
both NUGGET pricing and strategy, the NUGGET anthologies contain rich
labeled data for analy sis and conversion to an ill-defined neural
network compo nent.

2.1. Data Collection

To acquire sufficient diversified samples for our task, we conducted a
large scale user study. To avoid paying partic ipants, we relied on
good will (Friendship, 1901) and the unsubstantiated claim that paying
participants would skew the accuracy and impartiality of the
scientific results.

Our geographically diverse dataset of NUGGET pricing activations comes
from multiple samples across 8 coun tries: 2 from Brazil, 3 from
Australia, 2 from the conti nental United States, 1 from Germany, 1
from Malaysia, 1 from Thailand, 1 from the United Kingdom², and 1
from Japan. All participants in the user study found one or more
instantiations of NUGGET during their search, though this might be a
result of sampling bias³.

2.2. Non-linear NUGGET pricing

Rational consumers would expect that the price of a box of NUGGET
should increase linearly (or sub-linearly) as the quantity of NUGGET
is increased. From both individual experiments in NUGGET acquisition
and from our user study however we found this to not consistently be
the case.

²The authors note that United Kingdom should be United Queendom
whilst within a queen's reign but note this is out of the scope of
this work.

³The authors would like to know how to handle sampling bi ases but
carefully note that statistics is rarely used in machine learning and
that the Monty Hall problem is still highly confronta tional,
suggesting all later forms of statistics must be equally con
frontational. That's induction, right? Ugh, wait, that's math :(

[57]{.underline}

The NUGGET Non Linear Piecewise Activation

We propose taking advantage of these naturally occurring
{width="2.6010279965004375in"
height="3.468028215223097in"}

non-linearities to power our activation functions and show

that heavily used existing activation functions, such as the

Rectified Linear Unit (ReLU), fit within this framework.

The ReLU activation, mathematically defined as

ReLU(x) = max(0, x)

represents the optimal NUGGET pricing as determined

by a rational consumer. The price of a box of NUGGET

should increase proportionally to the amount of NUGGET

received. The max is a result of consumers being unable

to return or resell any amount of NUGGET to the original

producer of the NUGGET box⁴.

Even this cursory analysis suggests that the ReLU function,

traditionally attributed to , should be attributed to Professor

Robert C. Baker, creator of the NUGGET. We feel this is

a grave oversight in the current neural network literature.

Our work suggests researchers have issues with maintain

ing and tracking long term literature depedendencies, po

tentially due to truncated backpropagation through time.

Motivated by this rediscovery, we investigate whether other non-linear
NUGGET activations may act as a catalyst for the training and
production of neutral neural networks when subjected to a generative
adversarial setting⁵.

In Table 1 and 2, we explore non-linear pricing for a NUGGET box in San
Francisco, United States, for both McDonalds and Burger King (or Hungry
Jacks in Aus tralian). Note the price per NUGGET unit fluctuates wildly
between $0.149 and ∞.

3. Experiments

3.1. The Neural Numerology dataset

The Neural Numerology (MNIST) dataset contains 60,000 labeled images
of digits used to specify the quantity of a given NUGGET box.

Subjects were not required to make sensible orders, result ing in orders
of a zero NUGGET box and none where the NUGGET quantity exceeded nine.
Future work will rec tify this and allow for NUGGET boxes of ten to
twenty.

⁴The authors attempted multiple times to resell uneaten NUGGET
quantities to various fast food retailers. None of the initial trials
resulted in success and all subsequent attempts were met with a denial
of service (i.e. we were asked to leave the store).

⁵The authors do note that The Matrix (1999) can be seen as a
non-continuous generative adversarial multi-agent simulation. In
following work (Animatrix (2003), Reloaded (2003), Revolutions (2003)),
experimentation on humans in this manner was deemed unethical. We note
that the ethical treatment of neural networks when subjected to
adversarial settings has not yet been thoroughly discussed in the
literature but opt to ignore this insight by pretend ing this troubling
question had never been raised in the first place.

Figure 1. An architectural neuronal visualization produced when using
the NUGGET activation is substantially more aesthetic than that of
non-NUGGET based activation functions. Note the absence of killer
robots or glowing red eyes.

Nuggets Om nom Dollary-doos NUGGET unit

α = 0 ∅ $0.00 ∞ α = 4 XX $1.00 $0.25 α = 6 X $4.30 $0.72 α = 10 X
$4.99 $0.499 α = 20 XXXX $5.00 $0.25

Table 1. Non-linear NUGGET pricing at a McDonalds located in
contintental United States. At one extreme, increasing NUGGET quantity
by 2 results in $1.65 per NUGGET unit (4 → 6). At the other extreme,
increasing NUGGET quantity by 10 results in $0.001 per NUGGET unit
(10 → 20).

Nuggets Om nom Dollary-doos NUGGET unit

α = 0 ∅ $0.00 ∞ α = 10 XX $1.49 $0.149 α = 20 X $5.99 $0.299

Table 2. Non-linear NUGGET pricing at a Burger King located in
contintental United States. Note two n = 10 NUGGET boxes is cheaper
than an n = 20 NUGGET box. We are uncertain if gold or other valuable
items are in the n = 20 NUGGET box.

[58]{.underline}

The NUGGET Non Linear Piecewise Activation

That's pretty darn good. Few animals can read numbers or
{width="1.5532228783902011in"
height="1.5532228783902011in"}{width="1.553180227471566in"
height="1.553180227471566in"}

order nuggets, so our model is also smarter than most ani

mals and evolution took forever making those things.

5. Conclusion

In this work, we revisit the ReLU activation under the

framework of NUGGET based non-linear piecewise equa

tions. The improvements that these techniques provide can

likely be combined with other regularization techniques,

Figure 2. (Left) Neural Numerology samples generated without NUGGET
activations. (Right) Neural Numerology samples gen erated with
NUGGET activations. Notice the zeroes (0) have similar topology to
that of a traditional NUGGET blob.

3.2. Experimental setup

All experiments are implemented in PyTorch and are built upon existing
codebases. The use of existing code is essen tial as researchers are
still investigating how to make digital neurons feel warm and fuzzy
⁶. We elect not to use weight or batch normalization as the authors
are concerned with negatively impacting the neural network's body
image. For the same reason, we avoid using L₁ or L₂
regularization.

We considered using the Hogwild lock-free approach to parallelizing
stochastic gradient descent but elected against it as hogs are not
operationally equivalent to chickens and thus may invalidate our
results.

The neural network models were trained by a person named Adam
Optimizer and used an NVIDIA Volta whilst it was mining for Ethereum.
The learning rate began at 20 and was divided each time the training
curator Adam desired a NUGGET box of quantity one or more. This was
frequent.

All embedding weights were uniformly initialized in the interval [−0.1,
0.1] and all other weights were initialized between [−
^{[√1]{.underline}}_H, ^{[√1]{.underline}}_H], where H is the hidden
size. Anyone who guessed what the hidden size was won a prize.

4. Results

Our results ... are not that bad. Like, if you hired a five year old
to read the numbers in Figure 2 for you, that kid would probably do
worse than our algorithm. Therefore, NUGGET based artificially
intelligent models are equiva lent in complexity to that of a standard
human five year old.

⁶Many neural network experiments require dozens or hun dreds of
expensive high end GPUs, resulting in both massive ex pense and massive
heat generation. This is necessary as it helps incubate the neural
networks during their growth, with the GPUs helping heat them to their
optimal temperature (i.e. acting as a catalyst) and the dollar figure
spent on them ensuring the neural networks are aware of how much we love
them.

such as the drunken dropout, and may lead to further im provements in
performance as well, especially if subjected to an extensive global
NUGGET hyperparameter search. We see artisanal hand crafted activation
and regularization techniques the future of our field, primarily as
no-one is quite certain how a neural nets anyway.

Acknowledgements

We thank Charlie Yang for funding an experimental pur chase of an n =
20 NUGGET box that motivated this work. Additional NUGGET funders have
opted to remain anonymous due to the contentious nature of
artificially in telligent fast food research. Thanks to the
participants in the geographical NUGGET sampling: Anton Troynikov,
Joseph Stephen, Dominic Balasuriya, Georgina Wilcox, James Foster,
Joshua Hall, Kenya Chan, Dominick Ng, and Vivian Li. Good research not
only takes time and resources but also good friends. The authors would
perform better work if they had more friends. Please be our friend.

NUGGET samples

Sydney: 3 for $3, 6 for $6, 10 for $7.50, 20 for $12.75 Sydney
CBD: 3 for $3, 6 for $5.90, 10 for $7.70, 20 for $12.80

Melbourne: 3 for $3, 6 for $5.50, 10 for $7.20, 20 for $12.80, 24
for $9.95

Japan: 5 for 200 yen, 15 for 570 yen

UK: 6 for 3.09, 9 for 3.99, 20 for 4.99

Thailand: 6 for 87B, 10 for 139B, 20 for 240B Kuala Lumpur: 6 for
7.8RM, 9 for 10.9RM, 20 for 22RM Germany: 6 for e3,59, 9 for e4,49, 20
for e7,59 Belo Horizonte: 4 for 6.50 reais, 10 for 16.40 reais So
Paulo: 4 for 6.50 reais, 10 for 13.90 reais

US (McDonalds): see Table 1

US (Burger King): see Table 2

References

Srivastava, Nitish, Hinton, Geoffrey E., Krizhevsky, Alex, Sutskever,
Ilya, and Salakhutdinov, Ruslan. Dropout: a simple way to prevent
neural networks from overfitting. Journal of Machine Learning
Research, 15:1929--1958, 2014.

[59]{.underline}

Under review as a conference paper at SIGBOVIK 2018

SUBSTITUTE TEACHER NETWORKS:

LEARNING WITH ALMOST NO SUPERVISION 12

Samuel Albanie^∗

British Institute of Learning, Yearning and Discerning

Shelfanger, UK

James Thewlis^∗

National Academy of Pseudosciences

Valencia, Spain

Joao F. Henriques ˜^∗

Fortress of Solitude

Coimbra, Portugal

ABSTRACT

Education is expensive. Nowhere is that statement more universally
agreed upon than in machine learning, a recently trending topic on
twitter that places great value on the reduction of cost. Certainly
for machines to learn, they must be taught, but how can this be
achieved on an appropriate budget? Recent approaches (often referred
to as Teacher-Student or Knowledge Distillation methods in the
neural network literature) have demonstrated that the problem can be
viewed as model compression, in which a single student model learns
from an ensemble of M specialist consultants networks. Inspired by the
logo on a free pen at a local recruitment fair, we scale this method
up and out, while simultaneously pursuing an appropriately
aggressive patenting strategy. In total, we make the following three
contributions. First, we propose a novel almost no supervision
training algo rithm that is highly scalable in the number of student
networks being supervised. Second, we explore the closely-related
scaling problem of culinary optimisation, developing a method that
tastily surpasses the current state of the art. Finally, we provide a
rigorous quantitive analysis of our method, proving that we have
access to a calculator.

A little learning is a dangerous thing

Alexander Pope, 1709

1 INTRODUCTION

Since time immemorial, learning has been the foundation of Human
culture, allowing us to trick other animals into being our food. The
importance of teaching in Ancient Times was exemplified by Pythagoras,
who boasted of being able to teach his Theorem to anyone in the street
(Philolaus of Croton, 421 BC), though apparently no one taught him to
wear pants.

Nowadays, we are attempting to pass on this knowledge to our species'
offspring, the ma chines (Timberlake, 2028; JT-9000, 2029)¹, who
will hopefully keep us around to help with house chores.

^∗Authors listed in order of the number of guinea pigs they have
successfully taught to play competitive bridge. Ties are broken
alphabetically.

¹The work of these esteemed scholars indicates the imminent arrival
of general Artificial Intelligence. Their methodology consists of
advising haters, who might be inclined to say that it is fake, to take
note that it is in fact so real. The current authors, not having a
hateful disposition, take these claims at face value.

1

60

Under review as a conference paper at SIGBOVIK 2018

Many prominent figures of our time, several of whom cannot tell their
CIFAR-10 from their CIFAR 100 have expressed their reservations with
this approach, but really, what can possibly go wrong?² Moreover,
several prominent figures in our paper say otherwise (Fig. 1, Fig. 2).

Having established the wisdom of our approach as a whole with the
extensive philosophical discus sion above, we now press on to achieve
a finer understanding of the details. Concretely, the goal of this
work is to reduce the algorithmic ignorance, or more precisely
gnorance³ of a collection of student networks, and to do so in a
fiscally responsible manner given a fixed teaching budget.

Define a collection of teachers {T_e} as a class of highly educated
functions which efficiently map unusual life experiences residing a
Banach space into extremely unfair exam questions in an exami nation
space. Further, define a collection of students {S_t} as class of
keen beans which inefficiently map unheated pot noodles to unwashed
dishes, both in common space. Pioneering educational early work by
Bucilua et al. (2006) demonstrated that on a carefully illuminated
manifold, an arbitrary student S_t could improve his/her performance
with N highly experienced, specialist teachers. We refer to this as
the private tuition learning model. While effective in certain
settings, this approach does not scale. Specifically, this algorithm
scales in cost as O($MNK), where N is the number of students, M is
the number of private tutors per student and $K is price the bastards
charge per hour. Our key observation is that there is cheaper approach
to ignorance reduction, which we detail in Sec. 3.

Our work is biologically inspired by the humble ostrich, an animal
keenly aware of the dangers of learning too much, as its sand-based
defence mechanism affords it a heightened inability to perceive
threats. Advanced incomprehension of object permanence (Piaget, 1970)
is also a key characteristic of human infants, as demonstrated
empirically in the Stanford Peekaboo Experiment. This mental
peculiarity is even more pronounced in certain human adults, with
entire systems of contradictory beliefs able to be held simultaneously
and without distress. Similarly, a profound ignorance of neuroscience
allows the authors to confidently claim that the proposed method to
cost reduction during teaching is identical to neural pathways found
in the brain.

2 RELATED WORK

Give a student a fish and you feed them for

day, teach a student to gatecrash seminars

and you feed them until the day they move

to Google.

Andrew Ng, 2012

A worrying trend in the commoditization of education is the use of
MOOC (Massive Open Online Courses) by large internet companies. They
routinely train thousands of student networks in parallel with
different hyperparameters, some of whom are hurled out to the far east
on the explore-exploit coordinate chart, then keep only the
top-performer of the class (Li et al., 2016; Snoek et al., 2012). We
consider such practices to be wasteful and are totally not jealous at
all of their impressive com putational resources.

A number of approaches have been proposed to improve teaching quality.
Central to each of these approaches is a question that has challenged
researchers for many years, namely how best to ef ficiently extract
extract knowledge that is in the computer (Zoolander, 2004). Work by
noted en tomologists Dean, Hinton and Vinyals illustrated the benefits
of comfortable warmth to facilitate students better extracting
information from their teachers (Hinton et al., 2015). In more detail,
they advocated adjusting the value T in the softmax distribution:

p_{i =}exp (x_i/T)

[P]{.underline}

_jexp (x_j/T)(1)

²This question is rhetorical, and should be safe to ignore until the
Ampere release.

³The etymology of gnorance is a long and interesting one. Phonetic
experts will know that the g is silent (cf. the silent k in
knowledge), while legal experts will be aware that the preceding i
is conventionally dropped to avoid costly legal battles with the
widely feared litigation team of Apple Inc.

²61

Under review as a conference paper at SIGBOVIK 2018

Figure 1: We introduce Latent Substitute Teacher Allocation, a
simple generative process that ex plains the cost of learning. Note
the use of drop-shadow plate notation, which indicates the direction
of the nearest light source.

where T denotes the wattage of the classroom storage heater. More
radical approaches have advo cated the use of alcohol in the
classroom, something that we do not condone directly, although we
think it shows the right kind of attitude to innovation in education
(Crowley et al., 2017). However, both approaches are clearly
financially unsustainable. Moreover, differently from these works, we
focus on the quantity, rather than the quality of our teaching method.

Recent work has promoted an "Attend, Infer, Repeat" (Eslami et al.,
2016) approach to learning. Attendance is a prerequisite for our
model, and cases of truancy will be reported to the headmistress (see
Fig 1). For the substitute teacher module, the "Infer" step may be
replaced by "Ignore". Only particularly badly behaved student networks
will be required to repeat the course.

A number of pioneering ideas in scalable learning were physically
investigated several years ago by (Maturana & Fouhey, 2013). However,
we differentiate ourselves from their approach by using several orders
of magnitude fewer hashtags. We also note the marginal relevance of a
recent paper on unadversarial learning (Albanie et al., 2017). We now
attempt to cite a future paper, from which we shall cite the current
paper, in an ambitious attempt to send google scholar into an infinite
depth recursion (Albanie et al., 2019), thereby increasing our
academic credibility and assuredly landing us lucrative pension
schemes.

2.1 UNRELATED WORK

• A letter to the citizens of Pennsylvania on the necessity of
promoting agriculture, manufac tures, and the useful arts. George
Logan, 1800

• Claude Debussy---The Complete Works. Warner Music Group. 2017

• Article IV Consultation---Staff Report; Public Information Notice on
the Executive Board Discussion; and Statement by the Executive
Director for the Republic of Uzbekistan. IMF, 2008

• A treatise on the culture of peach trees. To which is added, a
treatise on the management of bees; and the improved treatment of
them. Thomas Wildman. 1768

3 THE LATENT SUBSTITUTE TEACHER ALLOCATION PROCESS

The primary goal of educators is to educate, inform and explain. In
machine learning, explana tions are best encoded as simple statistical
generative models. We therefore explain the role of cost efficient
explanation through an appropriately simple explanation, the Latent
Substitute Teacher Al location (see Fig. 1).

³62

Under review as a conference paper at SIGBOVIK 2018

80 60 40 20 0

---

---

0 20 40 60 80

Figure 2: Expressing the cost function in bitcoins makes it
significantly more volatile, yet it was instrumental in attracting
venture capital for our Smart Education startup.

Fortunately, since the model is graphical, it needs minimal
explanation. However, we can all agree that it will scale
magnificently. All the teacher networks employed in the Latent
Substitute Teacher Allocation Process are Recursive Neural Networks. A
Recursive Neural Network is defined as the composition of some layers,
and a Recursive Neural Network. By logical induction, these networks
have infinite capacity, which is why they are not bothered by a heavy
workload. All students are trained in two stages, separated by
puberty.

In keeping with the cost-cutting focus, we have analysed the gradients
available on the market, and after extensive research decided to use
Synthetic Gradients Jaderberg et al. (2016), which are significantly
cheaper than Natural Gradients Amari (1998). It is important to
realise that our cost function, which is the target of minimisation,
is very much proportional to actual cost (preferably cash; see Fig.
2).

Traditional approaches have often gone by the mantra that it takes a
village to raise a child. We attempted to use a village to train our
networks, but found it to be an expensive use of parish re sources,
and instead opted for the NVIDIA GTX 1080 Ti ProGamer-RGB. Installed
under a desk in the office, this setup provided warmth during the cold
winter months.

4 THE CAKE

As promised in the mouth watering abstract (and yet undelivered by the
paper so far), we now take a short, mid-paper confectionary diversion
to improve our ratings with the sweet-toothed de mographic⁴. A
number of competitive cakes have been recently proposed at a high-end
cooking workshop (LeCun, 2016; Abbeel, 2017), resulting in a dramatic
bake-off (Fig. 3-a,b).

Previous authors have focused on cherry-count. We show that better
results can be achieved with more layers, without resorting to
cherry-picking. Our layer cake consists of more layers than any
previous cake (Fig. 3-c), showcasing the depth of our work.

We would like to dive deep into the technical details of our novel use
of the No Free Lunch Theorem, Indian Buffet Processes and a
Slow-Mixing Markov Blender, but we feel that increasingly thin
culinary analogies are part of what's wrong with contemporary Machine
Learning (Rahimi, 2017).

⁴This approach was recommended by our marketing team, who told us
that everyone likes cake. ⁴63

Under review as a conference paper at SIGBOVIK 2018

Figure 3: Several cakes of importance for current research (deeper is
better). From left to right: 1) Yann LeCun's cake, 2) Pieter Abbeel's
cake, 3) Our cake. Note the abundance of layers in the latter.

5 EXPERIMENTS

If you don't know how to explain MNIST

to your LeNet, then you don't understand

digits!

Albert Einstein

We now rigorously evaluate the efficacy of the Latent Substitute
Teacher Allocation Process. We note that unlike previous methods, we
achieve regularisation without injecting gradient noise. High noise
levels tend to stop concentration gradients in student networks, and
learning stalls. In these experiments we always operate in
"library-mode". Performance-inducing drugs, such as batch norm, were
strictly prohibited.

After months of intensive training using our trusty NVIDIA
desk-warmer, which we were able to compress down to two days using
montage techniques and an 80's cassette of Survivor's "Eye of the
Tiger", our student networks were ready for action. The only
appropriate challenge for such well-trained networks, who eat digits
for breakfast, was to pass the Turing test. We thus embarked on a
journey to find out whether this test was even appropriate.

The Chinese Room argument, proposed by Searle (1980) in his landmark
paper about the philosophy of AI, provides a counterpoint. It is
claimed that an appropriately monolingual person in a room, equipped
with paper, pencil, and a rulebook on how to respond politely to any
written question in Chinese (by mapping appropriate input and output
symbols), would appear from the outside to speak Chinese, while the
person in the room would not actually understand the language. We ran
this thought experiment many times using the highly scalable nature of
the Latent Substitute Teacher Allocation Process. By sampling rulebook
operators appropriately from the earth's surface, we achieved strong
statistical guarantees that at least one of the monolingual subjects
would be appropriately Chinese. Having resolved all philosophical and
teleological questions, we then turned to the application of the
actual Turing tests.

Analysing the results in Table 4, we see that only the ResNet-50 got a
smiley face. The Q-network's low performance is obviously caused by
the fact that it plays too many Atari games. However, we note that it
could improve by spending less time on the Q's and more time on the
A's. The Neural

Model Turing test result

AlexNet B

ResNet-50 A+

Q-network C

Neural Turing Machine F-, see me after class

Figure 4: Results for the test class of 2018.

⁵64

Under review as a conference paper at SIGBOVIK 2018

Turing Machine had an abysmal score, which we later understood was
because it focused on an entirely different Turing concept.

As an additional, purely empirical statement, we observed that
networks trained using our method experience a much lower DropOut
rate. Some researchers set a DropOut rate of 50%, which we feel is
unnecessarily harsh on the student networks⁵.

6 CONCLUSION

You take the blue pill---the story ends, you

wake up in your bed and believe whatever

you want to believe. You take the red

pill---you stay in Wonderland, and I show

you how deep the ResNets go.

Kaiming He, 2015

This work has shown that it possible to achieve low-cost machine
learning by using inexpensive, completely expendable Substitute
Teacher Networks, while carefully avoiding their definition. We have
seen that residual networks may be the architecture of choice for
solving the Turing test. A major finding of this work, found during
cake consumption, is that current networks have a Long Short-Term
Memory, but they also have a Short Long-Term Memory. The permutations
of Short Short and Long-Long are left for future work, possibly in the
short-term, but probably in the long term.

ACKNOWLEDGEMENTS

This work was actively undermined by a wilful ignorance of related
work.

REFERENCES

Abbeel, Pieter. Keynote Address: Deep Learning for Robotics. 2017.

Albanie, Samuel, Ehrhardt, Sebastien, and Henriques, Jo ´ ao F.
Stopping gan violence: Generative ˜ unadversarial networks.
Proceedings of the 11^th ACH SIGBOVIK Special Interest Group on Harry
Quechua Bovik., 2017.

Albanie, Samuel, Ehrhardt, Sebastien, Thewlis, James, and Henriques,
Jo ´ ao F. Defeating google ˜ scholar with citations into the future.
Proceedings of the 13^th ACH SIGBOVIK Special Interest Group on Harry
Quechua Bovik., 2019.

Amari, Shun-Ichi. Natural gradient works efficiently in learning.
Neural computation, 10(2):251-- 276, 1998.

Amazon. Details redacted due to active NDA clause.

Bucilua, Cristian, Caruana, Rich, and Niculescu-Mizil, Alexandru.
Model compression. In Pro ceedings of the 12^th ACM SIGKDD
international conference on Knowledge discovery and data mining, pp.
535--541. ACM, 2006.

Crowley, Elliot J, Gray, Gavin, and Storkey, Amos. Moonshine:
Distilling with cheap convolutions. arXiv preprint arXiv:1711.02613,
2017.

Eslami, SM Ali, Heess, Nicolas, Weber, Theophane, Tassa, Yuval,
Szepesvari, David, Hinton, Geof frey E, et al. Attend, infer, repeat:
Fast scene understanding with generative models. In Advances in
Neural Information Processing Systems, pp. 3225--3233, 2016.

Hinton, Geoffrey, Vinyals, Oriol, and Dean, Jeff. Distilling the
knowledge in a neural network. In Neural Information Processing
Systems, conference on, 2015.

⁵This technique, often referred to in the business management
literature as Rank-and-Yank (Amazon), may be of limited effectiveness
in the classroom.

⁶65

Under review as a conference paper at SIGBOVIK 2018

Jaderberg, Max, Czarnecki, Wojciech Marian, Osindero, Simon, Vinyals,
Oriol, Graves, Alex, and Kavukcuoglu, Koray. Decoupled neural
interfaces using synthetic gradients. arXiv preprint
arXiv:1608.05343, 2016.

JT-9000. How I learned to stop worrying and love the machines
(Official Music Video). In British Machine Vision Conference (BMVC),
2029. West Butterwick, just 7 miles from Scunthorpe, England.

LeCun, Yann. Keynote Address: Predictive Learning. 2016.

Li, Lisha, Jamieson, Kevin, DeSalvo, Giulia, Rostamizadeh, Afshin, and
Talwalkar, Ameet. Hyperband: A novel bandit-based approach to
hyperparameter optimization. arXiv preprint arXiv:1603.06560, 2016.

Maturana, Daniel and Fouhey, David. You Only Learn Once - A
Stochastically Weighted AGGRe gation approach to online regret
minimization. In Proceedings of the 7^th ACH SIGBOVIK Special Interest
Group on Harry Quechua Bovik., 2013.

Philolaus of Croton. How to bake the perfect croton. Greek Journal of
Fine, Fine Cuisine, 421 BC. Piaget, Jean. Piaget's theory. 1970.

Rahimi, Ali. Test of Time Award Ceremony. 2017.

Searle, John R. Minds, brains, and programs. Behavioral and brain
sciences, 3(3):417--424, 1980.

Snoek, Jasper, Larochelle, Hugo, and Adams, Ryan P. Practical bayesian
optimization of machine learning algorithms. In Advances in neural
information processing systems, pp. 2951--2959, 2012.

Timberlake, Justin. Filthy (Official Music Video). In Pan-Asian Deep
Learning Conference, 2028. Kuala Lumpur, Malaysia.

Zoolander, Derek et. al. Zoolander. Paramount Pictures, 2004.

⁷66

CONFIDENTIAL COMMITTEE MATERIALS
{width="1.000485564304462in"
height="1.000485564304462in"}

SIGBOVIK 2018 Paper Review

Paper 28: Substitute Teacher Networks

Ben Blum, Light Cone Sedentarian

Rating: Defer

Confidence: Righteous

As to the authors' perspective I no doubt have already written, but to
my own have yet to write, in my review of (Albanie et al., 2019), the
use of forward citations to one's own future work is an irresponsible
act which degrades the fabric of academic space-time. I cannot condone
this practice and recommend the paper's publication be deferred until
2020.

Reviewer Two, Association for Confectionery Heresy

Rating: Accept

Confidence: Just here for the cake

The paper makes important advances in the area of research paper
structure. Specifically, the unrelated work section helps ground the
reader by providing a sense of the scope of the work, the use of
inspirational quotes is inspiring, and the use of the intermission
section (first proposed in SIGBOVIK Track L by [R. Two, 2015]) helps
to keep the reader's attention. I look forward to future work from
these authors on incorporating those concepts into their teaching
network itself to confer the same benefit upon students.

67

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

That's Numberwangcoin! Great guess---keep going!

switch (choose_dear_reader()) {

case 0: goto PAGE_51;

case 1: goto PAGE_51;

case 6: goto PAGE_51;

case 13: goto PAGE_51;

case 17: goto PAGE_51;

case e^π : goto PAGE_51;

case 42: goto PAGE_51;

case A_LOT: goto PAGE_68;

case ^{[−b±]{.underline}√}[b]{.underline}²[−4ac]{.underline}

_2a: goto PAGE_51;

case 1337: goto PAGE_51;

case 9001: goto PAGE_51;

case ONE_MILLION_DOLLARS: goto PAGE_51;

case 123456789101112: goto PAGE_51;

case ACKERMANN_5: goto PAGE_51;

case ℵ₁: goto PAGE_51;

}

68

SIGBOVIK 2018 {width="0.7500010936132984in"
height="0.7500021872265967in"}

(Continued) Message from the Organizing Commi ee

"Whoooo! Go robot, go!"

{width="4.277777777777778in"
height="1.3472222222222223in"}

switch (choose_dear_reader()) {

case FANTASTIC:

Press the down, left, right, and down pads in that order.

goto PAGE_87;

case DECENT:

Press the up, down, right, and right pads in that order.

goto PAGE_50;

case WAY_OFF:

Press the up, up, down, and down pads in that order.

goto PAGE_40;

}

69

70

Parapsychology

Get Out of My Head

13 This grad student studied parapsychology---and you won't believe
what he found

David Edelstein

Keywords: parapsychology, science, experimental design, experimenter
effects, philosophy, telepathy

[71]{.underline}

This Grad Student Studied 13

Parapsychology --- And You Won't Believe What He Found! David
Edelstein

Parapsychology is a scientific field studying effects that would be
extremely important to our understanding of the world, but are widely
considered to be nonexistent. As an attempt to examine the scientific
method, I conducted a parapsychology experiment to see if I could

telepathically influence people's minds. Participants were given
thirty seconds in which to click a tally counter while I either
mentally impelled them to push it more or did not. Analysis of the
results found that people in the control group pushed the button
statistically significantly more

frequently than people subject to the treatment condition --- I am
supernaturally unpersuasive. I consider several possible explanations
for this effect, including experimenter influences, failure of
blinding, coincidence, and actual telepathic faculty on my own part. I
discuss the implications of this experiment on my personal belief in
psi and on my attitudes towards science as a tool for uncovering the
truth.

[72]{.underline}

Introduction

Parapsychology is the study of psychic phenomena, of mental
capabilities beyond those explained by science. These are collectively
referred to as psi. Many topics fall within its umbrella; the ones
most relevant to this research are: [4]

• Telepathy --- Transmission, reception, and influence of thoughts

• Psychokinesis --- Exertion of physical force through mental power

• Precognition --- Divining inaccessible information about the future

Parapsychology promises revolutionary insights with incredible
applications. Telepathy could allow for rapid and surreptitious
communication, precognition could open new categories of computation
and math, and remote viewing even attracted CIA attention with Project
Star Gate [10]. New forces might be discovered, and a full science
of parapsychology would touch nearly every other discipline.
Personally, as a magician, I would be fascinated to learn about a real
version of the abilities I present only the facsimile of possessing.

There's extensive research into parapsychology by major academics such
as Daryl Bem, and it's published in dedicated peer-reviewed
parapsychology journals, with forays into top conventional psychology
ones [8]. However, parapsychology is overwhelmingly considered
pseudoscientific. Its claimed results tend not to replicate, its
mechanisms are hazy and frequently in contravention of existing
beliefs about physics, and its papers are often plagued by statistical
malpractice. Plainly, it's a field dedicated to studying an effect
that isn't real.

I agree. I am skeptical of psi (mostly; I confess to having
occasionally idly tried to move objects with my mind). However, I
think that the methodological criticisms of the field of
parapsychology also apply within the domain of conventional science.
Consider the replication crisis in social psychology, in which
statistical fraud has produced a wealth of groundless scholarship. One
article considers parapsychology as a control group of sorts for
science, studying a domain where there is no effect [2]. That they
nonetheless produce positive and negative results at a similar rate to
scientists in legitimate fields paints a concerning picture of the
ability of scientists to find affirmative outcomes if and only if
there is a true effect.

Felix Planer, the author of a book on esoteric beliefs, writes that
"[I]f the existence of PK [psychokinesis] had to be taken
seriously [...] no experiment could be relied upon to furnish
objective results, since all measurements would become falsified to a
greater or lesser degree, according to his PK ability, by the
experimenter\'s wishes." [1] He needn't have referenced
psychokinesis --- experimenter effects are a well-established
phenomenon in which the results of an experiment tend to be biased
towards those favored by the researcher [5]. Science doesn't need
psi to have a problem.

That's why I'm researching parapsychology. Science is a method, not a
domain, and I want to conduct an experiment in a strange domain
following scientific protocols to observe how they function and how
well I am able to follow them. That there is likely no true effect is
to my advantage, because it allows me to focus my analysis more on the
procedural elements of science. And if I do get positive results, that
will be all the more interesting.

Experimental Design

Am I able to telepathically influence people's behavior?

This is the question I will be researching, and to find out, I have
developed a protocol to isolate a causal effect between my thinking a
command at someone and their obeying it. I want this

[73]{.underline}

procedure to be quick, to pose no risks to the subjects or to myself,
to produce data more granular than a binary did or did not follow, and
provide little room for experimenter influence or mechanisms of causal
effect other than psi. At a high level, my procedure is as follows:

1. Greet and explain that the subject will have thirty seconds to
click a tally counter as many or as few times as they like, and that
interval will start and end with a musical note

2. I give them the tally counter and get out of their sight

3. I start the experiment sound file, randomly assign the subject,
and record their assignment

4. A note cues the subject that the test period has begun

5. Treatment conditions

1. Control: I read whatever's on my computer screen for thirty
seconds

2. Experimental: I intently think at the subject a psychic command to
push the button for the next thirty seconds, while avoiding giving any
audible indicators of my mental focus

6. A note cues the subject and myself to stop

7. The subject leaves and I record the number of times the counter
was clicked

Recruitment

For this experiment, I recruited almost exclusively people at Carnegie
Mellon University. They were primarily students, though several
subjects were professors, and some may have been merely on campus but
presently unaffiliated with the university. This was a highly educated
and likely unusually intelligent class of subjects. I did not record
demographic data, though from memory, participants skewed young ---
indeed, I don't believe any of them were over fifty --- and were
disproportionately Chinese and Indian, representative of the
demographics of Master's students at Carnegie Mellon. I believe gender
balance was fairly even.

Certain esoteric concerns regarding recruitment are also worth noting.
I did not ask my subjects about their attitudes towards psi, even
though it is frequently postulated in parapsychological research that
skepticism towards psi can inhibit its functioning [3].
Additionally, most people who took part in the experiment were
acquaintances or strangers. Few were friends and none were among the
people closest to me. If telepathic connection is possible, it may be
more readily established between minds already tuned to the same
frequency, as it were [7]. These both, then, could be significant
covariates.

I attempted to get as many people as possible to take part in the
experiment in the time I had, and so opportunistically recruited most
heavily from among my classmates. I was unable to conduct the
experiment until I had established the procedure and until I had
settled on a method and obtained the necessary materials, so my
recruitment was laggardly at the beginning. It also slowed on the last
few days of data collection, as I relaxed without the risk of having a
critically small sample size and exhausted the supply of willing
classmates. In total, I had 49 subjects.

Materials

I used a tally counter as shown in Figure 1 to be a button that
records the number of times it is pressed. I had considered
implementing a digital counter, but the mechanical device required no
technical implementation, had less risk of error either in function or
by the user, and would not tie up my computer during testing. The
tally counter proved perfect for the task.

I also needed a way to mark the bounds of a 30-second interval while
first having a short delay in which I could conduct my randomization.
For this, I composed a short track on Apple

[74]{.underline}

{width="2.4455971128608924in"
height="3.0291951006124234in"}Figure 1: The tally counter

GarageBand consisting of a piano note at 15 seconds, another identical
piano note at 45 seconds, and silence otherwise. In retrospect, I should
have had different sounds, and found a way to play these from a source
other than my computer. The inadequacy of this method of cuing the start
and end of the interval resulted in a handful of mangled data points.

Instructions

I did not have a fixed script for introducing subjects to the experiment
and giving them instructions, and it evolved over the course of data
gathering, though more or less attained fixity at something close to the
following:

I'm conducting an experiment that will just take a minute of your time.
This here is a tally counter. You push this button to increment it, and
the thing along the side here resets it --- you don't use that; I'll do
that after we're done. [Give tally counter to the participant.] In
this experiment, after a brief delay, you will hear a noise like this.
[Play sound.] That marks the start of a thirty second interval which
ends when you hear the same note. During

that interval, but only during that interval, you may press the button
to increment the tally counter as many or as few times as you like. At
the end, I'll recollect the tally counter and mark down how many times
you pressed it. For experimental integrity, you'll need to sit so that
you can't see me. Do you have any questions? Are you ready?

Even once my instructions standardized, there was still variation in
subjects' knowledge going into the experiment. Some subjects were
aware of the purpose of the experiment, something I had explicitly
mentioned in earlier versions of the instructions, and some had
watched the experiment being conducted on previous subjects.

Randomization

Proper randomization is vital for this experiment to control for
pre-randomization inconsistencies in the experimental procedures. This
ensures that many covariates are not confounders. Inconsistencies in
the instructions and in knowledge of the experiment, having taken
effect prior to random group assignment, could not have had causal
influence on my results. Similarly, I can rule out effect from the
room I conducted the experimented in (though I did record that), the
time it was conducted, individual variation in test subjects such as
their belief in psi, and many other factors.

Additionally, randomization also reduces the avenues for experimenter
effects. I could often tell whether a subject was likely to push the
button a large or small number of times. For instance, one subject
asked me what was the most number of times someone had clicked the
button, and unsurprisingly she recorded a very high number of presses.
If I did not randomize properly, it would be easy to produce whatever
results I desired. I also designed my procedure to randomize late, as
close to the effect administration interval as possible, so as to
limit my post randomization degrees of freedom.

To conduct my randomization, I used [random.org]{.underline}, which
promises high quality random number generation. I wanted to randomize
in a way that I knew I would not allow for experimenter effect. In
particular, I needed a method that would also minimize the risk of
esoteric influence on my part. A coin toss, for instance, can be
somewhat controlled by some magician's tricks

[75]{.underline}

(though not ones I myself have any practice at), and because this
experiment is examining psi, I also can't rule out the possibility of
telekinetic influence on the coin, similar to what Felix Planer feared
[1]. A website seemed sufficiently out of my control to be
trustworthy for these purposes.

Blinding

This experiment was single-blind. Double-blinding would have been
impossible, seeing as intention on the part of the experimenter is the
key independent variable. The protection of single-blinding varied
from subject to subject. For the first several subjects, I conducted
the experiment in a room with a screen I could stand behind out of
view of the subject. However, in later iterations, no such barrier was
available, and I had to make do with merely crouching out of sight
(hopefully) behind the subjects. Because the sound cues for the
interval played from my computer, where I also carried out the
randomization, I was unable to be very far away from the test subject
without risking them missing the interval. Most subjects never learned
which group they were in, and those that did learned only after their
data was collected and recorded.

Effect Administration

For subjects in the treatment group, I thought at them as hard as I
was able to over the thirty second interval a command to push the
button. Because I am not sure of the proper way to do this, my method
of application was inconsistent. For a significant majority of cases,
I simply thought in words "push the button" over and over for thirty
seconds. It varied whether I had my eyes open or closed for this, and
how much I breathed. I pushed in several different directions: that
the button itself was appealing, that the act of pushing it was
appealing, or that pushing it would be for science and demonstrate
something really cool. In some cases, I also mimed clicking the tally
counter, as though there were some synchrony between my body and that
of the subject.

I don't know which of these methods would be most effective, and I did
not record which one I followed, nor did I record the subjective
intensity of my treatment. I believe that in all cases, I was able to
deliver sustained and intense mental focus towards encouraging the
subject to push the button.

For subjects in the control group, I simply didn't think at them. To
best ensure that I would not accidentally go against the intended
treatment of the control group, I read. There was some variation in my
handling of the control group as well, as I would varyingly read from
a book or from what was on my screen, or examine other random things
generated by [random.org]{.underline}. It is possible that stray
thoughts towards pushing the button may have entered my mind while
some members of the control group were participating in the
experiment, but I can say with confidence that such thoughts were
never intense.

Data and Analysis

In total, I collected data from 49 people, reported in full in
Appendix A. Of them, 29 were in the control group and 20 were in the
treatment group. I chose to analyze the data using R because of my
prior experience with it and its extensive free statistical packages.
Across all participants,

the median number of clicks was 60, and the mean was about 58.7.
Standard deviation was 47.7. Two people pushed the button only once
over the thirty-second interval, and one person managed 159 pushes,
more than five per second.

Figure 2 shows box plots of the number of clicks by condition. The
control group had a mean of 72.3 and a standard deviation of 50.8,
whereas the treatment group had a mean of 40.0 and a standard
deviation of 35.4. The control group was fairly symmetrically and
widely distributed, but the treatment group was right-skewed, with
most of the data points below 30 and a long

[76]{.underline}