Sigbovik 2008

Click for full version PDF

A Record of The Proceedings of SIGBOVIK 2008

April 6^th, 2008

Carnegie Mellon University

Pittsburgh, Pennsylvania USA

http://sigbovik.org/

1

2

A Message From The Organizing Committee:

The Association for Computational Heresy Special Interest Group ACH
SIGBOVIK on Harry Q. Bovik is particularly excited to be presenting
this, the Second Annual Intercalary Workshop about Symposium on Robot
Dance Party of Conference in Celebration of Harry Q. Bovik s 2^{6 th}
Birthday. The previous Binarennial Scheduling of the SIGBOVIK confer

ences has been seen as a game changing development in the storied
history of conference presentation ever since George P. Burdell s
historic keynote talk at the rst SIGBOVIK in 1944 on the occasion of
Harry s 2^{0 st} birthday.

However, the response to what was is generally accepted as the seventh
SIGBOVIK Con ference¹ in 2007 was so overwhelming that the ACH
SIGBOVIK Governing Board was forced to recognize that it would be
simply negligent to allow the crucial work that nds unique expression
at SIGBOVIK to lie dormant until 2071. Therefore, the Intercalary
Workshops were announced, and this, this Second Intercalary Workshop
in Celebration of Harry s 2^{6 th} Birthday, is the rst of such
intercalary workshops that will help to advance the progress of
science until 2039, when the Thirty Second Intercalary Workshop in
Celebra tion of Harry s 2^{6 th} Birthday will be held concurrently
with the First Intercalary Workshop in Expectation of Harry s 2^7
th Birthday.

We hope that you will nd yourself edi ed and enlightened by this, the
proceedings of SIGBOVIK 2008. We, the pseudonymous SIGBOVIK 2008
Organizing Committee, are proud to present it, and we thank QVT
Financial LP, who boldly went where no corporation has gone before: to
sponsorship of SIGBOVIK.

Sincerely,

The SIGBOVIK 2008 Organizing Committee:

Ciel Elf General Chair

Guy Fantastic Program Chair

Emcee M.C. Assistant Program Chair

Tux Rat Publicity Co Chair

Forbes Anchovie Publicity Co Chair

Emmet O Theorem Design Chair

Julia Cette Pre Program Chair

Turing T. Turing Procurement Chair

Alga Rhythm Webmaster Chair

Alf A. Berry Local Arrangements Chair

Reginald Red Acted External Relations Chair

¹ The second and third SIGBOVIK conferences technically did not
happen due to Harry s approximately six year phase from early 1945 to
late 1950, and the seventh SIGBOVIK conference on Harry s 2^{6 th}
birthday was confusingly introduced in the proceedings as the sixth,
as the organizers counted from one instead of zero.

The SIGBOVIK Committee For Fixing The Numerical Errors In Last Year s
Introduction

3

4

TABLE OF CONTENTS

CHUCK NORRIS 7 Chuck Norris 9

COMPLEXITY THEORY 11 Apocalyptic Complexity 13 Paradoxical
Complexity 17

LIES, DAMN LIES, AND APPLICATIONS 19 Lies and the Afterlife 21
Lies and Caffeine 29 Lies and Disaster 31

GRATUITOUS INSULTS 35 You're a Jerk 37 No, You're a Jerk 41 Well,
You're a Slacker 45

SOFTWARE ENGINEERING 47 Engineering Zombie Survival 49 Engineering
Relentlessly 51 Engineering Mind Control 53

OMG NATURAL LANGUAGE LOL 57 Restoring Language 59 Eadday
Anguageslay 63 I HAS A LANGUAGE 75 Wiki Language 79

LOGIC AND APPLICATIONS 83 Objectivist Logic 85 Focused Logic 91
Heavy Metal Umlaut Logic 93 Classical Music Logic 95

REAL WORLD/COMPUTER INTERACTION 97 Operating Systems in the Real
World 99 Automatic Citations in the Real World 101 Censor Networks in
the Real World 105 Bipedal Robots in the Real World 107 Theory in the
Real World 111

5

6

Track 1:

Chuck Norris

Chuck Norris 9 Dinitz, Michael. An Incentive-Compatible Mechanism
for all Settings: Chuck Norris.

7

8

An Incentive-Compatible Mechanism for all Settings: Chuck Norris

Michael Dinitz

Computer Science Department

Carnegie Mellon University

mdinitz@cs.cmu.edu

April 6, 2008

Abstract

A very important property for any mechanism is
incentive-compatibility. A mechanism is incentive-compatible if no
agent has an incentive not to follow the protocol. In this paper we
present a new meta-mechanism that can be applied to any existing
mechanism to make it incentive compatible. This meta-mechanism is
Chuck Norris. This shows that any mechanism can be made
incentive-compatible, and thus the field of mechanism design is now
solved.

1 Introduction

Mechanism design is the attempt to design protocols where each agent
has their own selfish goals and is rationally attempting to optimize
them. This selfish rationality may result in agents refusing to
participate if they cannot benefit, or if they participate they might
lie or refuse to follow the protocol in some other way in order to
maximize their utility. The goal is to design mechanisms that are
incentive compatible, in which every agent has no incentive to deviate
from the specified protocol. We would also like our mechanism to
maximize the social welfare, usually defined as the sum of the
utilities of all of the agents. In many cases it is difficult to
develop mechanisms that are incentive-compatible and social welfare
maximizing, as sometime maximizing the social welfare will involve
punishing one agent in order to make the others happy, and thus this
one agent will not be incentivized to participate or to follow the
protocol. For a more in-depth introduction to the field of algorithmic
mechanism design and it motivations, see [1].

In this paper we prove the existence of incentive-compatible
mechanisms in all settings. We do this by constructing a
meta-mechanism that can be used to transform any existing mechanism to
make it incentive-compatible. This meta-mechanism can be described in
two words: Chuck Norris.

2 Main Result

Suppose that there are agents x₁, . . . , x_n, and the possible
outcomes of the protocol are in some set S. Each agent x_i has a
utility function u_{i : S → R. For every s ∈ S, let v(S) =}
�ⁿ_i=1 u_i(s) be the value (i.e. the social welfare) of the
solution. Let OPT ∈ S be the optimal solution, so OPT = argmax_s∈S
v(s). We say that a mechanism is an α-approximation if it returns a
solution s ∈ S such that v(s) ≥ v(OPT)/α.

Suppose there is some mechanism A which, if all agents follow the
mechanism, is an α-approximation. Let A^CHUCK be the following
mechanism. First, any agent that does not participate gets a visit
from Chuck Norris, who then proceeds to roundhouse kick the agent. We
then proceed according to A, but

9

any time an agent interacts with another agent or with the mechanism
Chuck Norris roundhouse kicks them if they do not follow the protocol.

Theorem 2.1 A^CHUCK is incentive-compatible.

Proof: A Chuck Norris-delivered roundhouse kick is the preferred
method of execution in 16 states [2]. Thus the utility to an agent
of any solution which involves being roundhouse kicked by Chuck Norris
is −∞, since that is the utility of death. It is easy to see from the
definition of A^CHUCK that any deviation from A by an agent will
result in a Chuck Norris roundhouse kick, and hence a utility of −∞.
So all agents will follow A, and thus A^CHUCK is
incentive-compatible.

The following corollary is almost immediate:

Corollary 2.2 A^CHUCK is an α-approximation

Proof: Recall that A is an α-approximation if all agents follow the
protocol. Since A^CHUCK is incentive-compatible we know that all
agents will follow the protocol. And except for possible Chuck Norris
roundhouse kicks A^CHUCK follows A exactly, so A^CHUCK is also an
α-approximation.

3 Discussion

In this section we discuss possible objections to the Chuck Norris
meta-mechanism. One possible problem is synchronous actions: if
multiple agents are all taking actions at the same time, then they all
have to be threatened by Chuck Norris, not just one of them. This is
not a problem, though, since a little-known (but very useful) folk
theorem states that "Contrary to popular belief, there is indeed
enough Chuck Norris to go around" [2]. A related objection is that,
even if Chuck Norris is physically able to administer a roundhouse
kick, non-compliance with the protocol might involve simply
misreporting private information, and thus Chuck Norris would not be
able to determine whether or not the protocol was followed. But this
is false, since Chuck Norris has the ability to read minds [2].

Finally, there is the possible issue of the utility of Chuck Norris
himself. After all, we crucially depend on his roundhouse kicks, and
while he obviously has the ability to roundhouse kick whomever he
wants, he might not have the desire. Fortunately an examination of the
other agents makes it clear that Chuck Norrs would indeed derive
utility from administering roundhouse kicks to the bad agents. This
follows from the fact that any agent which does not follow the
protocol has decided to ignore the threat of a Chuck Norris roundhouse
kick. This is obviously a foolish thing to do, and while "Mr. T pities
the fool, Chuck Norris roundhouse kicks the fool's head off" [2].

4 Conclusion

In this paper we have shown that any mechanism can be made
incentive-compatible by using Chuck Norris. This essentially solved
all open problems in the field of algorithmic mechanism design. Thus
Chuck Norris can add "solving all problems in algorithmic mechanism
design" to his formidable list of accomplishments.

References

[1] N. Nisan and A. Ronen. Algorithmic mechanism design (extended
abstract). In STOC '99: Proceedings of the thirty-first annual ACM
symposium on Theory of computing, pages 129--140, New York, NY, USA,
1999. ACM.

[2] C. Norris. http://www.chucknorrisfacts.com/.

10

Track 2:

Complexity Theory

Apocalyptic Complexity 13 Murhpy VII, Tom. non-non-destructive
strategy for proving P = NP.

Paradoxical Complexity 17 Klionsky, David. O(0) Algorithms and
Other

Applications of Time Travel.

11

12

A non-non-destructive strategy for proving P = NP

Tom Murphy VII

1 April 2008

Abstract

We provide a radical new approach for proving that P = NP, demonstrating
that if you put your mind to it, you can accomplish anything!

Keywords: complexity theory, p, np, prongs, hyper driven devices

1 Introduction

The field of Computers Science has been fairly suc cessful in answering
its "grand challenge" questions. For example, in 19XX Computers Science
answered in the affirmative, "Can a computer beat a human in Chess?" In
20XX we successfully built a Windows Vista. In 20XX Computers Science
answered in the affirmative, "Can a computer not be beaten by all humans
in Checkers?" However, some problems are still unsolved. Most vexing of
these is the question of P ^?= NP [Cook(1971)], that is, are
nondetermin istic Turing Machines inherently more efficient than
deterministic ones?

This is troublesome for a number of reasons. First, the existence of
unsolved problems adversely affects our "batting average," the primary
means for com paring Computers Science to other fields of import. (This
also has an indirect effect on other compara tive statistics, such as
the Earned Run Average of competing fields.) Such tarpits also waste
countless hours of ambitious graduate students's most creative years,
and the time and patience of program commit tee members for second- and
third-rate conferences. The open proposition also induces additional
market volatility, as futures markets¹ and institutions such

^-1Copyright c 2008 No Computers In Space LLC. Appears in SIGBOVIK
2008 with the permission of the Association for Computational Heresy;
IEEEEEE! press, Verlag-Verlag vol ume no. 0x41-2A. £0.00

¹Although no longer posing any risk to investors, the stillborn
ACM--NASCAR crossover Turing Machine 0x500 debacle---in which "races"
complete with pace rabbits were

as the Clay Mathematics Institute have placed boun ties totaling $1m
USD on its resolution---in either direction [Cook(2000)]. One is
also subject to the eerie suspicion that some hyperintelligent
observer is chuckling at our Sisyphean impotence, our endless attempts
at the same dead-end strategies and the puppy-like faithfulness with
which we return to the problem and continue to sing its praises.
Time's up. Put your pencils down and pass your papers to the front of
the class; the problem must be solved now.

In this paper we present a new strategy for prov ing that P = NP. This
approach differs from those that came before it in methodology and
consequences: It is inherently non-constructive (indeed, non-non
destructive) for one, meaning that we cannot directly use it as an
effective means for solving difficult (NP hard) problems in polynomial
time. In fact, the re sult makes the computational landscape less
efficient in general.

To begin, let us refresh our memories as to the statement of the P
^?= NP problem so that we can attack it where it is most weak.

2 Problem statement

The set of languages P is defined as fol lows [Cook(2000)].

P = { | = L(M)} for some Turing Machine M which runs in polyno

mial time

where is a language and L(M) is the language ac cepted by the machine
M.

staged between different Turing Machine programs solving var ious
problems with unknown complexity bounds, and holiday Vegas bettors
would have their pensions cleaned out by compu tational savant
bookmakers in smoky but mostly empty par lors designed to resemble
mainframe machine rooms---could also have been avoided had P ^?= NP
been solved prior to its inception.

13

Similarly, the set of languages NP is defined as

NP = { | = L(N)} for some non deterministic Turing

Machine N which runs

in polynomial time

where a non-deterministic machine is defined in the usual way.

Proving (or disproving) that deterministic and non deterministic
machines describe the same set of lan guages (by, for example,
establishing a polynomial time solution to an NP-hard problem, or giving
a lower bound for one) is famously difficult. In this paper we take a
completely different approach. The key observation is the implicit
existential quantifier in the definitions of P and NP: A language is in
P if there exists a polynomial time machine M such that = L(M). We
present a multi-pronged attack on existence by metaphysical arguments,
non-non destructive techniques, and complexity class mobil ity.

3 Do any Turing Machines ex ist?

The first question we can ask is: Do any Turing Ma chines actually
exist? If not, then the languages P and NP are empty, and trivially
equal. One can make a reasonable case that, in fact, there are no Tur
ing Machines; the machines require an infinite-length tape, an object
that many object to the existence of in the physical universe. (Some
argue that the lack of evidence for infinite tape is actually a planned
obso lescence conspiracy by the 3M corporation, and that infinite rolls
of tape are in fact present in their un derground laboratories.)

However, even if Turing Machines do not exist in the physical universe,
most Mathematicians and Computers Scientists would be prepared to accept
the existence of Turing Machines within the Platonic universe of
idealized mathematical objects. Here, in finite tape is in abundant
supply. The Platonic uni verse fortunately also affords the ability to
carry out many other feats of the mind, which abilities we use in the
next prong.

4 Destroy all Turing Machines

Supposing that Turing Machines do already exist, and we find this to
not be desirable, we still do have

{width="2.926000656167979in"
height="3.2476662292213474in"}

Figure 1: An example of a multi-pronged attack on Turing Machines.

recourse.

For example, many systems for formal mathemat ics such as The C
Programming Language and L^AT_EX support the ability to remove or
alter definitions in the environment (for example through #undef and
\renewcommand). Why should not these constructs of human thought be
available to us in the Platonic universe? Specifically, why should we
not be able to make Turing Machines not exist by the power of human
thought alone? The traditionally non destructive nature of the
Platonic universe compels us to forever recall our inconvenient
mistakes. This is, frankly, some intolerable bullshit. Are Computers
Scientists ready to admit that their thoughts are not powerful enough
to undo their own other thoughts?

Effecting this change might not be so simple. The Platonic universe is
a mathematical commons shared by all clear thinkers. Observing our
weakness and our attempt to subvert it, competing fields may very well
cause Turing Machines to come back into exis tence by redefining them
to their current pernicious meaning. Maintaining a force of constantly
vigilant Computers Scientists to battle the existence of Tur ing
Machines could be as wasteful as attempting to decide P ^?= NP
through conventional means. In stead, we should use our creative
powers to populate the competitive idea landscape with countermeasures

14

Figure 2: A non-non-more-destructive approach. (a) is the normal
inclusion diagram for P and NP in the absence of an answer to P ^?= NP.
In (b), for each language in P, we forget all polynomial time
algorithms. This leaves only the exponential time solutions, making P =
NP ©.

to prevent Turing Machines without constant atten tion. For example, I
am currently imagining a mys tical boomerang-like five-pointed Glaive
weapon that has been rescued from a lava cave such as like in the 1983
heroic fantasy film Krull and which has the abil ity to chop up a Turing
Machine's big ol' tape like superheated tungsten piano wire through a
deciliter of I Can't Believe It's Not Butter brand butter-like spread.
This weapon I've imagined is chopping up Turing Machines at a rate of
like an Ω stack of Ωs every second, and I'm just getting started (Figure
1)! By populating the Platonic universe with such non non-destructive
thoughts, we can keep it essentially clean of working Turing Machines
and simultaneously produce Platonic block-buster films on the cheap.

5 Rise Up! Up the polynomial hierarchy!

On the other hand, many people have become rather attached to
computation and its useful fruits. What if we are unwilling to abandon
computation altogether? A second approach is inspired by the asymmetry
in the difficulty of the P = NP question: It is very easy to prove that
P ⊇ NP but difficult to prove that NP ⊇ P. Rather than refute the
existence of Turing Machines, we achieve P = NP by "forgetting" all of
the polynomial time algorithms for solving prob lems in P (Figure 2).
Since deterministic exponen tial time suffices for solving every problem
in NP (by exhaustive search), if we also have only exponen tial time
algorithms for solving problems in P, then these two complexity classes
will be equivalent.² As a

²PS. My Krull weapon flying around idea space is prevent ing you
from noticing that this argument does not make sense (Figure 1).

result, our existing computer programs will run more slowly, but we
will at least simultaneously be able to have computation and a
satisfactory solution to the vexing P ^?= NP problem.

6 Related work

Others have proposed trivializing solutions to the P ^?= NP problem,
such as the algebraic solutions N = 1 or P = 0. This is pretty dumb.

7 Conclusion

I hereby authorize the Clay Mathematics Institute to direct deposit
$1m USD into my bank account, routing number 7474-133-790.

References

[Cook(2000)] Stephen Cook. The P ver sus NP problem, May 2000. URL
http://www.claymath.org/millennium/P_
vs_NP/Official_Problem_Description.pdf. Clay Mathematics Institute
Millennium Prob lems.

[Cook(1971)] Stephen Cook. The complexity of theorem-proving
procedures. In Conference Record of Third Annual ACM Symposium on The
ory of Computing, pages 151--158, 1971.

15

16

O(0) Algorithms and Other Applications of Time Travel

David Klionsky

Carnegie Mellon University

April 1, 2008

Abstract

Asymptotic analysis has hit an asymptote in its ability to classify
algorithmic complexity. We propose a new order of functions, zero time
functions, to classify the set of functions that terminate before they
are run, using novel applications of age-old time travel techniques.
Other topics include the McFly Theorem (an extension of the Mas ter
Theorem), the Bill and Ted (BT) class of algorithms named for its most
excellent founders, and the Primer Conjecture which we are certain no
one understands. We also prove that P does indeed equal NP. We've been
to the future, people, just trust us on this one.

17

18

Track 3:

Lies, Damn Lies, and

Applications

Lies and The Afterlife 21 McGlohon, Mary and Robert J. Simmons.
Toward a Frequentist Approach to Pascal\'s Wager.

Lies and Caffeine 29 Landwehr, Peter, D Lee, Mary McGlohon, and
Rob Reeder. Dolla Bill Y\'all: Is There a Bias to the Orientation of
Dollar Bills Put in the Coke Machine?

Lies and Disaster 31 McGlohon, Mary. Data Mining Disasters: A
Report.

19

20

Towards a Frequentist's Approach

to Pascal's Wager

Mary McGlohon Robert J. Simmons^∗

April 6, 2008

Abstract

Pascal's wager attempts to provide a mortal with a proper choice of
believing or not believing in a god, based on the expected reward of a
given belief. It is essentially a Bayesian approach to the existence
of a supreme being, as it deals with a degree of belief approach to
proability. However, given the ineffability of a supreme being, the
idea of finding a Bayesian prior for performing inference is
impractical. However given the high population of observable mortals,
a frequency probability would be a more obvious choice. Therefore, we
present a systemized frequentist approach to the problem of a supreme
being.

{width="2.2276662292213474in"
height="1.3684995625546807in"}{width="2.2276662292213474in"
height="1.3684995625546807in"}

Figure 1: Diagramming the utility in situations where there is a
choice you have control over (&) and another choice you do not have
control over (⊗). On the left, bringing wine when fish or steak may be
served. On the right, believing in God when God may or may not exist.

21

1 Introduction

1.1 Pascal's Wager

We follow Giden Rosen's description of Pascal's Wager [8], also know
as Pascal's Gambit. First, we note that is possible to drawn a chart
that describes the different choices available to an actor on the
y-axis and the different possible states of the world (which are
assumed to be unknown to the actor) on the x-axis. As with most
problems in life, this problem can be recast in terms of alcohol
[9]; in particular a situation in which the actor has the option of
bringing red or white wine to a friends' house without knowing whether
chicken [7] or wine will be served for dinner.

As everyone knows [1], chicken with white wine is pretty good, but
chicken with red wine is so so, whereas steak with red wine is
freakin' amazing but white wine with steak is no good. By assiging a
numerical value to the utility of each of these combinations, we can
obtain the graph in Figure 1. A risk averse actor would be inclined to
bring red wine unless there was no possibility of steak being served;
however, the behavior of a fully rational actor will be determined by
the probabilities they assign to the different possibilities.
Presuming that there is an equal probability of either possibility,
then the expected utility of bringing white wine is 7 × .5 + −2 × .5 =
2.5, whereas the expected utility of bringing red wine is 3 × .5 + 11
× .5 = 7, and the rational actor will bring red wine, as the expected
gain for doing so is 4.5 units. On the other hand, if the rational
actor thinks that there is an 80% chance of chicken being served, then
the expected utility of bringing white wine is 7 × .8 + −2 × .2 = 5.2,
whereas the expected utility of bringing red wine is 3 × .8 + 11 × .2
= 4.6, and the rational actor will bring white wine, as the expected
gain for doing so is .6 units.

Pascal's Wager seeks to extend this ordinary and legitimate reasoning
to the case for belief in a deity. One version of the argument
imagines that there is an inherent utility of a human life, k, and
that the value of k + α is the value of a life lived acting under the
belief in the existance of God. Some versions of Pascal's Wager take α
to be positive, some nega tive, typically depending on how much people
like guilt and/or Gregorian chant. Then, posit that either no God
exists, or else there is a God who rewards His believers with eternal
bliss -- we will describe this God as a "rational rewarding" God.
Definitionally, we can assume that the utility of "eternal bliss" is
infinite, and seeing as k+α+∞ = ∞ as long as k and α are finite, we
end up with the chart on the right-hand side of Figure 1.

At first glance, we must expect an actor to assign some probability of
the existence of God, and some other probability to the non-existence
of God, and proceed by the same analysis we used for deciding whether
to bring wine. If we assign that God exists with probability p, then
the expected utility of non-belief is k × (1 − p) + k × p = k, and the
expected utility of belief (k+α)×(1−p)+∞×p = ∞, and so a rational
actor should believe in God, as the expected gain for doing so is a
rather persuasive ∞ units.

^∗The results contained herein reflect neither the opinions of the
authors, nor those of the National Science Foundation.

22

{width="4.685999562554681in"
height="1.3676662292213473in"}Figure 2: A philosopher-mathemetician's
critique of Pascal's Wager

1.2 A Mathematical Critique

The historical critique of Pascal's Wager, as described by Giden Rosen
[8], falls into two categories. The first is a theological critique;
in a world where multiple religions teach eternal punishment or reward
for belief/nonbelief in their god, Pascal's construciton gives
little-to-no guidance for the prob lem of picking "the right God."
This problem will not be considered in this paper due to restrictions
[4], and in any case, this critique of Pascal's Gambit is
well-understood.

A more basic mathematical critique begins with the idea that the
infinities in present in Figure 1 are suspect from a mathematical
point of view. We can drive this concern home by assigning non-zero
probability to a God which we call "perverse, active" and which Rosen
describes as philosopher-friendly. "I didn't give them any evidence of
existance," this God thinks, "and by golly, those non-believers, they
stuck to their guns. I'll give them eternal bliss, and give the
believers eternal punishment."

Now the non-believer has an expected utility of ∞, and the expected
utility of the believer is... one must suppose, impossible to
calculate. We can add even more absurdity to the Pascal argument by
positing the existence of a God that sends believers to heaven or hell
with probability .8 and .2, respectively, whilst leaving nonbelievers
alone. The analysis used in descriptions of Pascal's Wager becomes
completely inadequate in this environment, though one must assume in
such a universe non-belief and risk aversion would have to be linked.

1.3 A Frequentist Critique

Since the question of using Bayesian or frequentist approaches to sta
tistical analysis is a nearly religious debate in the field [5], the
obvious extension is to apply it to relgious matters. Furthermore, it
assumes the "gambling god" to be introduced later-- or, more
generally, a god that does not consider gambling a punishable sin.

On the other hand, we do have billions of observable mortals, so as
signing a frequency probability to the existence of a supreme being
would be a more natural way of going about things in the supernatural
realm.

23

2 Methodology

In order to determine an appropriately frequentist, we needed a sample
space of universes . We wanted to investigate a wide variety of possible
of potential God-models, including Gods that behave rationally
(consistently rewarding those that believe in them), perversely
(consistently punishing, or failing to reward, those that believe in
them), or arbitrarily (meting out eternal reward or punishment in a
manner that is only rational with some probability, which may or may not
be contgient on belief).

2.1 Sampling

2.1.1 Rapture-Recapture

We introduce a novel method of sampling for supernatural experiences,
which we call Rapture-Recapture. We first chose at random 100 people
from each of 6 universes: Earth, Bizarro, World of Warcraft, Star Trek,
Star Trek Mirror Universe, and the Buffyverse. We surveyed each subject
regarding their beliefs in god, humanity, and their own sins. We then
tagged the right ear of each subject and euthanized them. After some
period of time we performed a re-capture and again surveyed each re
captured subject on their posthumous experiences.

2.1.2 Entrance survey

Before euthanization, we presented each subject with an extensive survey
with questions regarding their faith, time spent on earth, and other nec
essary information to obtain before euthanization. The survey is
included in Appendix A.

2.1.3 Euthanization

We then attempt a re-capture through wireless transmission. As we as
sume that everyone in heaven gets a free iPhone, and everyone in hell
gets a Bluetooth Ouija Boards, we ensure that our hardware is compatible
with both.

Zombiefication was also used as a backup method of obtaining posthu mous
survey data. It was only used as a backup, as the IRB would not approve
the proposal to use zombiefication and revive people already in heaven.

2.1.4 Exit survey

Of the re-captured subjects, we obtained a completed survey from each,
shown in Appendix B.

3 Results

Results from some of the universes sampled are presented. 24

{width="2.2094444444444443in"
height="2.3634995625546806in"}

Figure 3: One of the huntards that nearly killed the undergrad we
hired to gather data from Azeroth. (picture courtesy of
www.figurerealm.com)

3.1 World of Warcraft

The World of Warcraft (WoW) universe, termed Azeroth, has a number of
interesting differences that often were an advantage for our
experiment. A resurrection (rez) system is in place, in which players
spend some amount of time essentially dead while their disembodied
soul has to run from the graveyard back to the place they were ganked;
this is known as a corpserun. During this time they are still able to
use voice chat to communicate, which made our devices described
earlier unnecessary.

Several difficulties arose in performing the rapture-recapture. It was
difficult to get an unbiased sample, as whenever we tried to use
subjects from parties, particulary pick-up-groups that included
paladins (sometimes priests and shamans, because those were usually
n00bs (or n00badins) with no respect for science and tended to
interfere by casting healing spells upon our subjects or prematurely
rezzing them. Secondly, on several occasions some huntard would sic
their pet, usually a tiger, on the experimenters (see Fig. 3).
Thirdly, occasionally warlocks stole the souls of dead characters and
captured them in soulstones. Since we considered that to the an
interruption of the normal rapture-recapture experiment, we were
unable to use those data.

Results were somewhat inconclusive. Despite the built-in ease of com
municating with un-rezzed characters, usually they went AFK (away from
keyboard), as if ordering pizza were more important than the progress
of science.

3.2 Bizarro World

The Bizarro World of Htrae functions in every way imaginable opposite
of planet Earth. Very pleasingly, we thereby found opposite results.
While

25

we inferred from exit surveys that 10% of earthly subjects went to
some version of eternal bliss, 90% of Bizarro subjects did.

4 Conclusion

We have not had time to fully analyze the results, and periodic de
monic posession by our Subversion server has been a constant source of
\<\<\< mine, all mine! bwahahahaha ==== >>> r666 ==== We are confi
dent that our data sets will be a useful for future study. Hey, we put
"Towards" in the title, didn't we?

References

[1] Like, duh.

[2] http://www.saintaquinas.com/mortal sin.html.

[3] http://www.icanhascheezburger.com.

[4] William Kristol. Oh, the anguish!: The cartoon jihad is phony.
The Weekly Standard, 11(22), February 2006.

[5] John Lafferty and Larry Wasserman. All of Statistical Machine
Learning. Pink Book Publishers, 2012.

[6] Jim McCann and Ronit Slyper. A theft-based approach to 3d object
acquisition. In SIGBVOIK, 2007.

[7] Mary McGlohon. Fried chicken bucket processes. In The 6^th Biaren
nial Workshop about Symposium on Robot Dance Party of Conference in
Celebration of Harry Q. Bovik's 0x40th Birthday, April 2007.

[8] Gideon Rosen. Pascal's wager, 2002. Lecture, Introduction to
Meta physics and Epistomology, Princeton University.

[9] Robert J. Simmons. A non-judgmental reconstruction of drunken
logic. In The 6^th Biarennial Workshop about Symposium on Robot Dance
Party of Conference in Celebration of Harry Q. Bovik's 0x40th
Birthday, April 2007.

[10] David Steiner. Proof-theroetic strength of pron with various
exten sions, 2001.

[11] Tom Murphy VII. Name of Author by Title of Book. Lulu Press,
2003.

[12] Tom Murphy VII, Tom Murphy VII, Tom Murphy VII, Tom Mur phy
VII, Tom Murphy VII, Tom Murphy VII, Tom Murphy VII, Tom Murphy VII,
and Tom Murphy VII. Level of detail typesetting in academic
publications. In SIGBOVIK, 2007.

26

APPENDIX

A Entrance survey

1. How many supreme beings do you believe in? (if less than one, skip
to Question 2)

$a$ Do they insist they are the only god(s)?

$b$ Do they insist upon belief in them for a good afterlife? ©
Very important for this study What do they say about re garding the
eternal fate of people dying through assisted suicide or otherwise
consenting to their own death?

2. Have you participated in a study like this before?

3. Have you experienced any death or near-death experiences? 4. Did
you commit any of the following? [2] Please estimate the number of
times. (If no exact count is known, please give a relative term such
as 'a few times', 'more than Larry King', 'did not inhale', etc.): (a)
Idolatry Includes sacrilege, sorcery

$b$ Pride Includes atheism, citing your own paper [11], © Lust
Includes adultery, fornication, prostitution, rape, sodomy incest,
masturbation, divorce, pornography, typesetting porn [12], kitty
porn [3], PRON [10],

$d$ Gluttony Includes over-consumption of food and alcohol, bad
table manners. See also idolatry of Ben and Jerry.

$e$ Sloth Includes observing the Sabbath, not observing the Sab
bath,

$f$ Greed Includes theft [6], perjury, fraud, extortion, usury,
more cowbell, saving a bundle on car insurance.

$g$ Wrath Includes murder, suicide, abortion, terrorism, Also in
cludes self-destructive behavior such as alcohol abuse, drug abuse,
and grad school.

$h$ Sins of fashion Includes blue eye shadow, Mom Jeans, dress ing
like a computer scientist, wearing white after Labor Day, shopping at
Ikea after completing a college degree.

$i$ Sins against animals Includes dog shows, eating meat, wear ing
leather.

$j$ Sins against humanity Includes being a jerk, using passive
voice, editing your own wikipedia article, off-color jokes, voting for
Ron Paul.

5. Please list any atonement you performed for acts in Question 4.

B Exit survey

1. Do you know you are dead?

2. What is your current quality of life, compared to your life on
earth? 3. What is the current temperature?

27

28

Dolla Dolla Bill Y'all: Is There a Bias to the Orientation of Dollar
Bills Put in the Coke Machine?

Peter Landwehr, D Lee, Mary

McGlohon, Rob Reeder

Carnegie Mellon University

5000 Forbes Ave.

Pittsburgh, PA, USA

{bovik, bovik, bovik, bovik}@cs.cmu.edu

ABSTRACT {width="2.501000656167979in"
height="1.279332895888014in"}

Yes.

Keywords

Coke, machine, dollar, yes

1. INTRODUCTION

Coke purchases are initiated by the input of cash into the Coke machine
[1]. This cash is often in a form of a US$1 (dollar) bill placed into
the bill slot on the front of the ma chine. When a dollar bill is placed
into this slot, the bill's front-to-back orientation must be face-up,
but its top-to bottom orientation can optionally be lefthand (i.e., with
the top of George Washington's head to the left) or righthand (i.e.,
with the top of George Washington's head to the right). The machine will
accept bills in either the lefthand or right hand orientation. See
Figure 1 for photographs of accepted orientations.

If the people placing bills into the machine pulled them out of their
pockets and wallets and placed them into the machine in an orientation
chosen at random, we would ex pect half the bills to be placed in the
machine to have the lefthand orientation and half to have the righthand
orienta tion.

We are led to our pressing research question: Are bills placed into
the Coke machine with a random orientation, or is there a bias toward
either the lefthand or righthand orientation?

2. METHODOLOGY

On two different days, we opened the Coke machine and carefully
removed dollar bills from the bill slot receptacle so as to preserve
their orientation. We then counted the num ber of bills in each
orientation. Bills are emptied from the receptacle at unpredictable
intervals by mysterious people,

Copyright is held by the author/owner. Permission to make digital or
hard copies of all or part of this work for personal or classroom use is
granted without fee.

SIGBOVIK (SIGBOVIK) 2008, April 6, 2008, Pittsburgh, PA USA .

Figure 1: Dollar bills shown in the lefthand (left) and righthand
(right) orientations, both of which are accepted by the Coke machine.
Both bills are shown in the face-up front-to-back orientation, which
is re quired by the machine.

but it can be safely assumed that those bills in the recepta cle at
the time of observation are a random sample of bills placed in the
machine for Coke purchase. Thus, while we were not sampling all bills
placed in the Coke machine, our sample is a fair random sample of all
bills.

3. RESULTS

Results of our study can be seen in Table 1. On both days of
observation, more bills were found in the lefthand orien tation than
the righthand orientation. The total counts over both days of
observation were 72 bills in the lefthand orien tation and 41 in the
righthand orientation. In percentages, this is 64% lefthand versus 36%
righthand.

Table 1: Number of dollar bills observed in the left hand orientation
and righthand orientation on two days of observation. The results show
a clear bias toward the lefthand orientation.

+-----------------+-----------------+----------------+----------------+
| | > Lefthand | > Righthand | > Total |
+=================+=================+================+================+
| > Day 1 | > 9 | > 7 | > 16 |
+-----------------+-----------------+----------------+----------------+
| > Day 2 | > 63 | > 34 | > 97 |
+-----------------+-----------------+----------------+----------------+
| > Total | > 72 | > 41 | > 113 |
+-----------------+-----------------+----------------+----------------+

To determine whether the observed lefthand-orientation bias was
statistically significant, we computed the probabil ity of observing
72 of 113 lefthand-oriented bills under the null hypothesis that no
bias exists. Under the null hypoth esis, lefthand bill orientation is
distributed binomially with a probability p of 0.5 and number of
observations n of 113,

{width="2.500999562554681in"
height="3.1909995625546808in"}

Figure 2: A sign on the Coke machine suggests that bills should be
placed into the slot in the lefthand orientation.

and the probability of observing at least 72 lefthand-oriented bills is
0.002, far less than the standard experimental alpha of 0.05. We thus
reject the null hypothesis and conclude that our result is strongly
statistically significant.

4. DISCUSSION

Our results provide strong evidence that people are biased toward
placing bills in the Coke machine in the lefthand orientation. There are
several possible explanations for this bias:

• A sign on the Coke machine bill slot suggests a left hand
orientation is the correct orientation for placing bills into the
slot, and does not suggest that any other orientation will be
accepted. This sign is pictured in Figure 2. Everyone likes to obey
posted signs. Except, apparently, for the 36% of people who ignore
this sign and righthand orient their bills.

• The natural orientation in which people store bills in their pockets
or wallets and people's natural physiol

ogy may be such that when they pull out bills to place them in the
Coke machine, the bills are more often lefthand-oriented. For example,
it may be that most people are righthanded and sort bills in their
wallets with the top of George Washington's head pointing up toward
the slit in the wallet, and that they open their wallets with their
right hands while pulling out the bills with their left and inserting
them into the machine in the lefthand orientation.

• In the lefthand orientation, George Washington is fac ing into the
bill slot. It may be that most people ap preciate the grim symbolism
of making our beloved Founding Father watch as he is gobbled up by the
vo racious corporate behemoth embodied in the Coke ma chine.

We suspect the sign on the machine is the primary cause of the
lefthand-orientation bias, but our study cannot distin guish the
effects of any one cause on bill orientation from any other cause. We
must leave it to future work to determine the cause of this unexpected
but important phenomenon.

5. CONCLUSION

Is there a bias to the orientation of dollar bills put in the Coke
machine? Yes. Lefthand.

6. FUTURE WORK

In future work, which will almost certainly never be done, we will
measure the effects of different possible explana tory causes on Coke
machine bill orientation. We will screw around with the directional
indicator sign on the machine and maybe some other variables until we
fully understand why people choose one orientation over the other.

7. ACKNOWLEDGEMENTS

We thank the Coke machine maintainers and Dec/5 trea surers who made
this work possible. That includes ourselves, whom we thank most
heartily. Think of us the next time you lefthand-orient your bill in
the Coke machine.

8. REFERENCES

[1] CMU SCS Coke Machine maintainers. CMU SCS Coke Machine. Web
page, March 2008. Available at http://www.cs.cmu.edu/~coke/. Accessed
on March 15, 2008.

Data Mining Disasters: a report

Mary McGlohon

Carnegie Mellon University

Machine Forgetting Department

5000 Forbes Ave.

Pittsburgh, Penn. USA

mmcgloho@cs.cmu.edu

{width="2.363500656167979in"
height="2.1876673228346455in"}{width="3.3218339895013123in"
height="2.3943339895013125in"}Figure 2: This is probably a log-normal
distribution.

This is not a power law.

Figure 1: ERROR::NumericOverflow. Nobody an

ticipated the breach of the levees.

1.2 Power law failures

ABSTRACT

Preventing data mining disasters is an important problem in ensuring
the profitability and safety of the field of data mining. Some data
mining disasters include decision tree forest fires, numerical
overflow, power law failure, danger ous BLASTing, and an associated
risk of voting fraud. This work surveys a number of data mining
disasters and pro poses several prevention techniques.

1. DATA MINING DISASTERS AND REC OMMENDATIONS

1.1 Numeric overflow

Numeric overflow is a significant problem in machine learn ing
programming. In 2007, numeric floods caused over $600 million in
property damages [1], and a loss of several thou sand nerd-hours of
work.¹ A lack of response fromthe Pro gramming Emergency Management
Agency (PEMA) was also often cited as an issue in such catastrophes.

When faced with a situation of numeric floods (such as that shown in
Fig. 1.1), a drowning researcher's best bet is to grab hold of a
floating log among the debris.

¹1 nerd-hour = 1 grad-student hour = 6 undergrad-hours = 0.5
faculty-hours

While much natural phenomena follow long-tailed distri butions, there
is a tendency to believe that everything is self-similar and that all
long-tailed distributions are equiva lent to power-laws (see Fig.
1.2). This has become a source of debate between computer scientists,
physicists, and statis ticians. The last group tends to be very
particular on what constitutes a "distribution". A debate may be found
in [3, 9].

Techniques for avoiding this sort of power-law failure are described
in detail in [4].

A possibly more dire form of power-law failure occurs when researchers
spend too much time arguing whether or not some long-tailed-looking
data actually comes from a power law, log-normal, or doubly-Pareto
log-normal gener ator. Everybody knows that things get nasty when
statis ticians get religious about something (for instance, the turf
wars between rapping statisticians Emcee M.C. and the Un biased M.L.E
[7]).

1.3 Decision tree forest fires

Occasionally researchers using pruning algorithms on their decision
trees get carried away. Instead of pruning unneces sary branches in
the interests of reducing overfitting. The experimenter just burns
down the tree until it is a decision stump. Repeating this on every
decision tree built is what is termed a decision tree forest fire (see
Fig. 3). This is not to

31

{width="3.150167322834646in"
height="3.2876662292213474in"}

Figure 3: Remember, kids, only you can prevent decision tree forest
fires.

be confused with the Forest Fire Model, a generative model for
evolving social networks [5].

As prevention measures, researchers should obtain a burn ing permit
before choosing to prune their decision trees with fire. Also, smoking
while researching is not recommended, and anyone engaging in such
behavior should ensure that their "butts are out".

1.4 BLAST accidents

Bioinformatic tool Basic Local Alignment Search Tool (BLAST) [2] is
useful for comparing sequences of amino-acids in pro teins, or of
base-pairs in DNA sequences. However, if used improperly, it can be
over-sensitive. This is what we term a mining BLAST accident.

A recommendation to avoid such disasters it for researchers to be
properly trained in using BLAST, as well as alternative algorithms for
subsequence matching.

1.5 Voting fraud by one-armed bandits Data mining also may suffer
cascading failures from er rors made in other fields. Two important
game theory and mechanism design subfields are voting mechanisms and
one armed bandit problems [10]. A fatal mistake is made when
combining the two, which results in inaccurate data; thereby creating
data mining disasters when data mining researchers attempt to use
these data.

There are several common methods that one-armed ban dits use of
committing voter fraud. For instance, they may impersonate actual voting
machines (see Fig. 4). They may also try to confuse polling officials by
citing various viola tions of policies set by the Americans with
Disabilities Act. They may also cram cake[6] into the voting
machines².

²The cake is a lie.

32

{width="1.969332895888014in"
height="2.578500656167979in"}

Figure 4: This is what happens when you don't pay attention in your
undergrad AI class.

{width="1.969332895888014in"
height="2.2651673228346456in"}

Figure 5: Regulation safety helmets for data miners can prevent
accidents.

2. OTHER PREVENTION TECHNIQUES

2.1 Cool Helmets

As a safety precaution, data miners should wear mining helmets, such
as that shown in Fig. 5. And overalls, ideally. This will also serve
to legitimize data mining as a real field of mining.³ As a result,
it will raise morale among researchers and prevent the often fatal
results of data mining accidents.

3. CONCLUSIONS

The author hopes that this paper will raise awareness among data
miners of risks involved in the field of practi cal prevention
techniques. When faced with any sort of data mining disaster, it is
generally advisable to remain calm and

³Talismans such as scarves, fanny packs, and pony-tails may also
serve as good-luck charms in preventing data mining disasters.

blame it on one-off errors, lack of rigor in proofs of correct ness,
or whatever government agency is funding the project.

Acknowledgments

Some images were borrowed from various sources on the Internet and
blatantly defiled with MS Paint. The original image used in Fig. 1.1
was provided by the Associated Press. The image for Fig. 3 was
borrowed from Tom Mitchell's web page for his textbook [8]. Sources
for Fig. 4 include digi talmedia.ucf.edu and www.thewe.cc. In Fig. 5,
Christos Faloutsos is modeling a mining helmet found at golden
westtravel.net.

4. REFERENCES

[1] Made up statistics, 2008.

[2] S. F. Altschul, W. Gish, W. Miller, E. W. Myers, and D. J.
Lipman. Basic local alignment search tool. J Mol Biol,
215(3):403--410, October 1990.

[3] A.-L. Barabasi. The origin of bursts and heavy tails in human
dynamics. Nature, 435:207, 2005.

[4] A. Clauset, C. Shalizi, and M. E. J. Newman. Power-law
distributions in empirical data. 2007.

[5] J. Leskovec, J. Kleinberg, and C. Faloutsos. Graphs over time:
densification laws, shrinking diameters and possible explanations. In
KDD '05: Proceeding of the eleventh ACM SIGKDD international
conference on Knowledge discovery in data mining, pages 177--187, New
York, NY, USA, 2005. ACM Press.

[6] M. Magdon-Ismail, C. Busch, and M. Krishnamoorthy. Cake-cutting
is not a piece of cake, 2002.

[7] M. McGlohon. Methods and uses of graph

demoralization. In The 6^th Biarennial Workshop about Symposium on
Robot Dance Party of Conference in Celebration of Harry Q. Bovik's
0x40th Birthday, Apr. 2007.

[8] T. Mitchell. Machine Learning. McGraw-Hill Education (ISE
Editions), October 1997.

[9] D. B. Stouffer, R. D. Malmgren, and L. A. N. Amaral. Comment on
barabasi, nature 435, 207 (2005). 2005. [10] M. Wooldridge.
Introduction to MultiAgent Systems. John Wiley & Sons, June 2002.

33

34

Track 4:

Gratuitous Insults

You\'re a jerk 37 Anchovie, Forbes. Maximum-jerk motion planning.

No, you\'re a jerk 41 Kua, John and Pras Velagapudi. Optimal Jerk

Trajectories.

Well, you\'re a slacker 45 Dinitz, Michael. Slacking with Slack.

35

36

Maximum-Jerk Motion Planning

Forbes Anchovie^∗

Carnegie Mellon University

Figure 1: Queuing is often corrupted by non-minimal-jerk actors. Here,
arriving A can choose between minimal jerk path B and more realistic
path C.

Abstract

Path planning is important for robot manipulators and other au tonomous
systems. There is strong evidence in the biomechanics literature to
suggest that smooth, natural, trajectories can be ob tained by a planner
which minimizes the fourth derivative of posi tion, or "jerk". In this
paper we present observations of behavior which seem to contradict this
biomechanical result. We use these as motivation to formulate a more
realistic path-planning paradigm based on maximizing the fourth
derivative of acceleration. These "maximum-jerk" trajectories are found
to accurately replicate ob served behavior.

CR Categories: X.2.3 [Activity Recognition]: Jerky Behavior---
Planning

Keywords: robot, motion planning, complete bastard
^∗e-mail:jmccann@cs.cmu.edu

1 Introduction

In many circumstances, smooth and pleasing trajectories for robotic
manipulators may be obtained through the use of the "minimum jerk"
criterion. Are these paths realistic? Certainly such trajectories
match well the measured human movements in a laboratory setting.
Outside of a laboratory, however, things are hardly that simple. It is
our observation that real-world trajectories are rarely as nice. In
light of this observation we have formulated a new planning model
which attempts to maximize jerk. In addition, we provide a some what
depressing theoretical result that shows that the jerk of some
trajectories can actually be unbounded.

This paper is organized as follows: In §1 we introduce the paper; in
§2 we gloss over previous work; in §3 we provide motivating
observations; in §4 we give example results; and in §5 we conclude.

37

$a$ (b) ©

Figure 2: Path examples in driving. In panel (A), hungry computer
scientists in car B wish to make a right turn; bus C is stopped,
blocking the lane. In panel (B), we show the minimum jerk path,
which would be to just let us into the traffic flow -- you're not
getting there any faster anyway. In panel (C), we show the
observed path, which was to just block us like a total bastard.

2 Background

The minimum-jerk trajectory model was both formulated and eval uated by
Flash ¹ and Hogan [1985]. They found that, in laboratory conditions,
the predictions of the minimum-jerk model matched well with measured
results for planar two-joint trajectories.

Their model is a straightforward minimization over trajectory
x(t): �

4 Results

In practice, observed trajectories closely match the theoretical max
imal jerk actions. In our driving experiences, illustrated in part in
Figure 2, we found that people are discourteous jackasses. Our re
sponding hand gesture -- see Figure 3 -- indicated our displeasure
and, to be candid, was far from being remotely minimal jerk. Our
planner suggested an alternate action, also pictured, which would have
been socially relevant. Arriving at our destination late, we

argmin*_x*

^˙x¨(t)² (1) t

were faced with another decision -- see Figure 1. Unfortunately, in this
case, the maximal jerk action proved infeasible.

This simple formulation lends itself to implementation. Previous work
has shown that minimum-jerk plans are useful in cooperative
manufacturing environments [2006]. Perhaps more surprisingly, a
minimum-jerk planner has been used to give people the robotic fin ger
[2004] (we also provide examples in this regime Figure 3).

3 Observations

We set out to study trajectories of people outside of laboratory con
ditions [MTV 2005]. We studied three standard conditions:

1. queuing [Zone 1998],

2. city traffic [Soderbergh 2000],

3. and restaurants [Veber 1998].

We performed our investigation by driving out to nice restaurant without
reservations, waiting to get in, then staring uncomfortably at the other
patrons until we were evicted from the premises. This kept our
experience under-budget. We recorded all observed behav iors on large
yellow legal pads using oversized novelty pens.

¹Flash -- a-ah -- savior of the universe! [May and Mercury 1980]

In addition to comparing our theoretical results, we have im plimented
a maximal-jerk controller for the Shadow Robot Hand [Laboratory
1999] (see Figure 4). Implimentation was simple once we overcame the
Shadow Hand's [Smith 1776] proper british upbringing.

4.1 Unbounded Jerk

In laboratory conditions we have been able to produce signals with
nearly unbounded jerk without notable visual distortion. We do this by
adding a rapidly-varying yet low-magnitude sine wave to a trajectory.

Starting with example point-to-point trajectory x(t) define x
(t) ≡ x(t) +ε sin(φ*t*) (2)

Notice that while the deviation from the path is proportional to ε the
additional jerk added to the path

^˙¨x _{(t) =} ^˙x¨(t)−εφ³ cos(φ*t*) (3)

is proportional to the cube of the frequency of the deviation. For an
example of this construction in practise, see Figure 5.

38

$a$ (b) © [Hitchcock 1963] (d)

Figure 3: Path examples for hand manipulator. In panel (A), the
hand is ready to signal after the events in Figure 2-(C). In panel
(B), the minimum jerk signal is a friendly wave; "we see you, next
time." In panel (C), the maximum jerk signal is less friendly. Our
planner occasionally was drawn to the local maximum shown in (D).

{width="1.5059995625546807in"
height="2.4959995625546805in"}{width="1.6918339895013124in"
height="2.4959995625546805in"}REDACTED
{width="1.289332895888014in"
height="2.503500656167979in"}(a) (b) © (d)

Figure 4: Hand trajectories demonstrated on the Shadow Hand robotic
hand platform.

Of course, in a real situation it is debatable whether the frequency φ
is actually unbounded.

5 Conclusions

In this paper we provided justification for the existence of a regime
of motion planning strategies that seek to maximize jerk. We rode this
justification to eventual sunset glory by creating plans that matched
the behavior of those real-world agents we observed. We additionally
provided a theoretical result that indicates that un bounded jerks may
appear entirely normal. This result is, to say the least, unsettling;
in the future it would be interesting to perform a survey to determine
what constitutes a just-noticeable jerk and use this to get a bound on
the maximum feasible asshole.

Acknowledgments

Thanks to the pile of money found in Wean hall grant, the Hugh H.
grant [Hefner 1953], and unbridled enthusiasm. We'd also like to
thank the academy. And ourselves, for putting up with our incessant
and improper use of the first-person plural.

References

FLASH, T., AND HOGAN, N. 1985. The coordination of arm move ments: an
experimentally confirmed mathematical model. Jour nal of Neuroscience
5, 7, 1688--1703.

GYORFI, J., AND WU, C.-H. 2006. A minimum-jerk speed planning
algorithm for coordinated planning and control of auto mated assembly
manufacturing. Automation Science and Engi neering, IEEE Transactions
on [see also Robotics and Automa tion, IEEE Transactions on] 3, 4,
454--462.

39

1.5 1

0.5 0

-0.5 -1

-1.5

x(t)

x'(t)

additional jerk

-1 -0.5 0 0.5 1

Figure 5: Unbounded jerk. The original and modified paths are so close
as to be indistinguishable, but the additional jerk (dotted line) is
substantial.

HEFNER, H. 1953. Playboy, vol. 1. Playboy Enterprises, Inc.,
December.

HITCHCOCK, A., 1963. The birds.

LABORATORY, H. 1999. Super Smash Bros. Nintendo.

MAY, B., AND MERCURY, F. 1980. Flash gordon (soundtrack). Hollywood
Records, 1.

MTV, 2005. The real world.

SECCO, E. L., VISIOLI, A., AND MAGENES, G. 2004. Minimum jerk motion
planning for a prosthetic finger. J. Robot. Syst. 21, 7, 361--368.

SMITH, A. 1776. The Wealth of Nations.

SODERBERGH, S., 2000. Traffic - die macht des kartells. VEBER, F.,
1998. Le dˆıner de con.

ZONE, V., 1998. John garwood. Starring Manolo Quequing as Young
Mishima.

40

Optimal Jerk Trajectories

John Kua and Pras Velagapudi

Institutionalized Robotics

Carnage Melon University

Pittsburgh, PA 15213

Email: {jkua, pkv}@cmu.edu

Abstract---Yeah. You'd like that, wouldn't you. A nice, short
{width="3.004763779527559in"
height="0.5959995625546807in"}

abstract so that you can just toss the rest of this paper in

the garbage. Just enough so that you can answer one or two

questions from your adviser about our approach, and then ignore

us forever. Well we won't have any part of it! You're going to have to
at least look at the captions and skim the introduction and
conclusion, you jerk!

I. INTRODUCTION

There are many examples in the field of minimal jerk trajectory planning
for robots [1] [2] and humans [3] [4]. However, these papers
labor under the assumption that robots and/or humans wish to minimize
jerkiness. We believe that this is not always the case - that under
certain scenarios, jerkiness is highly desirable, for instance, when
someone has seriously cheesed you off. However, jerkiness is not
directly related to energy expenditure. Increasing energy can increase
the jerk magnitude, however, this is not always the case. Indeed, we
posit that there is a bound on the maximum magnitude of jerk possible.
An example of this maximum jerk scenario is destroying the target's
property, domicile, and finally, the target. Potentially, one could
include destroying the target's home planet [5], but we believe the
target will no longer care. Indeed, we theorize that a target subjected
to constant levels of jerkiness will become inured, thus establishing an
upper bound.

As such, in this paper we propose that there are jerk trajectories which
are optimal. These trajectories are optimal in that they maximize the
jerk to energy ratio (JTE). Expending energy beyond this optimal level
is simply a waste of time and effort. We will show examples and analyses
of such optimal jerk trajectories.

II. ANALYSIS OF OPTIMAL JERK TRAJECTORIES

The illustrations in Figure 1, courtesy of [6], describe examples of
optimal jerk trajectories. In Figure 1(a), we see a very simple and
low energy optimal jerk trajectory (OJT). While the target is not
looking, one turns off the lights in the room and leaves. This is
highly irritating to the target, who must now fumble about the room
for the light switch. With one's departure, the target has no idea who
the culprit is. Hi-larious. Figure 1(b) is a very satisfying OJT, with
a very direct blow to the (presumably annoying) target's head, laying
them out on the floor. A very popular OJT with a wide variety of
results is the "seasoned" drink, shown in Figure 1©. The choice of
"seasoning" controls the outcome of this

$a$ Turn Lights Off

{width="3.003902012248469in"
height="0.596000656167979in"}(b) Punch

{width="3.003902012248469in"
height="0.5951662292213473in"}© Poison

{width="3.004763779527559in"
height="0.599332895888014in"}(d) Garotte

{width="3.0039173228346456in"
height="0.6034995625546806in"}(e) Shove

{width="3.004763779527559in"
height="0.599332895888014in"}(f) Dispose Body

Fig. 1. If you are doing these things, you might be a jerk.

OJT, ranging from unpleasant flavors to psychedelic drugs and
ultimately iocaine poison. The classic OJT is the garotte, shown in
Figure 1(d). Here one approaches the target from behind and strangles
them with a garotte, e.g. a length of fiber wire. The jerk factor is
quite high in this example, as the target experiences significant pain
before dying.

If it can be arranged, an excellent OJT is the staged accident. The
"stage" here is, for example, a high balcony where the target is
smoking a cigarette or a cliff edge as the target enjoys the view.
Then with a simple shove, the target falls to their

41

death, or at least a significant maiming. This is shown in Figure
1(e).

One simple method to maximize the jerk to energy ratio is disposing
the body in a dumpster afterwards, as shown in Figure 1(f). This hides
the body and allows it to decompose nicely before it is discovered.
This prevents the target from having an open casket funeral and gives
the bugs something nice to eat as well.

We can easily verify the optimality of such trajectories by applying
the following logical proof, derived with assistance from the
handwaving logic set forth by [7]:

{width="3.5070002187226597in"
height="1.4301662292213473in"}

It is clear that, through this exemplary triumph of modern proofery,
we can not only verify the optimality of our trajec tories, but also
save the whales.

III. COMPUTING OPTIMAL JERK TRAJECTORIES From these examples, it is
clear that a procedure is necessary to generate an OJT between
arbitrary start and goal points. We present the following completely
legitimate solution to the generalized OJT problem. Transform the
obstacles of the workspace into configuration space in closed form.
Reduce the dimensionality of the problem to a 2-D real-valued space by
eliminating stupid dimensions like left and up. Finally, map the
remaining configuration space to polar coordinates over a disc 18" in
diameter. The experimenter must then proceed to the nearest location
that provides alcoholic beverages and a dartboard. Locate at least 10
darts and attain a BAC of 0.08. Now close your eyes, spin exactly
500^◦, and throw a dart. Repeat this process for all darts, or until
physical violence ensues.

It has been shown that the problem of escaping a drunken brawl can be
reduced to any unconstrained OJT problem, thus the resulting escape
trajectory used by the experimenter will solve the OJT over the
original space. By using a radially constrained polar mapping, it is
ensured that as long as the experimenter travels at least 18", a
complete solution can be found. If they do not make it at least this
far, the solution will be incomplete, and the experimenter will really
hate the problem in the morning.

IV. REGIONS OF INEVITABLE JERKINESS

In many domains, computational effort may be saved by avoiding the
explicit computation of OJTs. Instead, environ ments can be broadly
decomposed into regions of inevitable jerkiness (ROIJ). Such regions
exist in almost any scenario, allowing near-optimal jerk trajectories
(NOJTs) to be formed

{width="3.0049311023622045in"
height="3.3726662292213474in"}

Fig. 2. The first Google Images result for "optimal jerk trajectory."
Incidentally, also the first Google Images result for "reaching to
grasp the apparatus," if you know what we mean. And we think you do.

{width="3.000999562554681in"
height="2.4509995625546805in"}

Fig. 3. A man, a large heavy block, and the corresponding regions of
inevitable jerkiness. Also, some sort of weird driving show on TV or
something.

by searching through possible motions through these regions. One
simple example can be seen in Figure 3.

Within regions, OJTs can be computed by transforming the problem to
its hyper-dual, the canonical homicidal chauffeur problem [8]. When
a solution is computed, it can either be transformed back to the
original problem and solved or, if a limo can be located, be directly
executed in its hyper-mega dual form. In an interesting special case,
both the original trajectory solution and its
pseudo-ultra-hyper-mega-dual can be proven to be OJTs.

42

{width="3.000999562554681in"
height="3.6334995625546807in"}Fig. 4. Human Tetris is fun.

V. CONCLUSION

The awesomeness of these methods may be able to be shown using the
handwaving logic set forth by [7], however, the authors feel that
this may not be strong enough, and will resort to Jedi mind tricks as
demonstrated in [9]. These trajectories are optimal. These aren't
the droids you're looking for. You may go about your business. Move
along.

ACKNOWLEDGMENTS

The authors would like to thank the cast of MTV's The Real World and
the Pennsylvania Department of Transportation for their continuing
production of jerk datasets. This work was partially supported by the
Internal Revenue Service grant

ISO-9000102-13892-TURTLE-9321309.1293929.

REFERENCES

[1] Piazzi, A. and Visioli, A., "Global minimum-jerk trajectory
planning of robot manipulators," Industrial Electronics, IEEE
Transactions on, vol. 47, no. 1, pp. 140-149, Feb 2000.

[2] Simon, D., "Application of neural networks to optimal robot
trajectory planning," Robotics and Autonomous Systems, vol. 11, no. 1,
p. 23-34, 1993.

[3] God, et al., The Holy Bible, nth ed., Heaven, ∞.

[4] Gautama, S., "The Teachings of Gautama Buddha," India, 500 B.C.
[5] Adams, D., The Hitch-Hiker's Guide to the Galaxy. Pan Books,
1979. London, England.

[6] Uncredited Artist, "The Professional's Methodology," Hitman:
Blood Money Documentation, Eidos Interactive, 2006.

[7] Simmons, R., "A non-judgemental reconstruction of drunken
logic," Proceedings from SIGBOVIK 2007, pp. 11-15, April 2007.

[8] Merz, A. W., "The homicidal chauffeur (pursuit-evasion
differential game)," AIAA Journal. Vol. 12, pp. 259, 260. Mar. 1974

[9] Lucas, G., et al., "Star Wars: Episode IV - A New Hope,"
Lucasfilm, 1977.

43

44

Slacking with Slack

Michael Dinitz

Computer Science Department

Carnegie Mellon University

mdinitz@cs.cmu.edu

April 6, 2008

Abstract

The classical graduate student problem is the well-studied problem of
how a graduate student can spend all of their time slacking off in
graduate school while still graduating. A famous impossibility result
of Bovik [3] states that if all of a student's time is spent
slacking, then it is impossible to graduate. We relax this problem by
adding a slack parameter , representing the fraction of time that the
student has to spend working. On this fraction we make no guarantee at
all about the enjoyment of the student, but this enables us to
guarantee graduation while also guaranteeing large enjoyment on the
other 1 − fraction of the time.

1 Introduction

It is well-established that the goal of graduate school is to slack
off as much as possible while still eventually graduating [6].
Unfortunately it is impossible to both slack off all of the time and
still graduate [3]. We can alternatively try for a more fine-grained
analysis, where there is an unhappiness level at every time and the
goal is to minimize the total unhappiness (the integral over time)
while still graduating, where the unhappiness is a function of the
current state (working or slacking) and the previous history of
states. Suppose that graduate school last for n years. It is known
that under plausible productivity and unhappiness functions, the
minimum amount of unhappiness required is still Ω(log n).

In order to get around this lower bound we introduce a slack parameter
. This slack parameter lets us ignore the unhappiness at an fraction
of the time (i.e. an n total amount of time). In other words, we get
to choose intervals of total length at most n and take unhappiness
integral over all times not in the segments. We show that by doing
this we can drastically decrease the unhappiness, from Ω(log n) to
O(log ^{[1]{.underline}} ). Thus if is a constant, we can get down to
constant unhappiness!

1.1 Related Work

In the last few years there has been a great deal of work on problems
with slack parameters. Slack was originally defined by Kleinberg,
Slivkins, and Wexler [7] in the context of metric embeddings. They
proved that by ignoring an fraction of the pairs in the metric space,
the distortion of the rest can be made extremely small. This was
continued by Abraham et al. in [1], and taken even further by
Abraham, Bartal, and Neiman [2]. It was first studied in contexts
other than metric embeddings by Chan, Dinitz, and Gupta [4], who
studied spanners with slack. Their techniques were then used by Dinitz
to give good compact routing schemes with slack [5].

45

2 Slack Construction

Our construction is based on the following simple observation:
graduate student unhappiness is sharply concentrated around a few
specific events. These events are the thesis defense, the thesis
proposal, the speaking skills talk, and advisor meetings, all of which
require considerable work and thus do not allow for significant
slacking off. But since these events together are only a negligible
fraction of the time that a student spends in graduate school, by
ignoring the unhappiness of these times we see a drastic decrease in
unhappiness. This is formalized by the following theorem:

Theorem 2.1 Let u : R⁺ → [0, 1] be an unhappiness function that is
O(1)-concentrated around the thesis defense, thesis proposal, and
advisor meetings, where u(t) = 1 means extreme unhappiness and u(t) =
0 means no unhappiness. Then there is a slacking schedule s : R⁺ →
{0, 1} (where 0 represents slacking and 1 represents working) and an
ignore function g : R⁺ → {0, 1} such that

� _n t=0

_{u(t)s(t)g(t)dt ≤ O(log} ¹ )

_{where g is only 1 on an fraction of the time, i.e.} �
ⁿ_t=0 g(t)dt ≤ n. Furthermore, at time n the student actually
manages to graduate.

Proof: Deferred to the full version, or left as an exercise for the
interested reader if the full version is never written.

3 Conclusion

We have proved that by enduring a few periods of extreme unhappiness,
it is possible to graduate with only mild total other unhappiness.
Yay!

References

[1] I. Abraham, Y. Bartal, T.-H. H. Chan, K. Dhamdhere, A. Gupta, J.
Kleinberg, O. Neiman, and A. Slivkins. Metric embeddings with relaxed
guarantees. In FOCS '05: Proceedings of the 46^th Annual IEEE Symposium
on Foundations of Computer Science, pages 83--100, Washington, DC,
USA, 2005. IEEE Computer Society.

[2] I. Abraham, Y. Bartal, and O. Neiman. Advances in metric
embedding theory. In 38^th STOC, 2006.

[3] H. Q. Bovik. Slacking, n-1 letters, and graduation rates. In
SIGBOVIK '07, 2007.

[4] T.-H. H. Chan, M. Dinitz, and A. Gupta. Spanners with slack. In
ESA'06: Proceedings of the 14^th Annual European Symposium on
Algorithms, pages 196--207, London, UK, 2006. Springer-Verlag.

[5] M. Dinitz. Compact routing with slack. In PODC '07: Proceedings
of the twenty-sixth annual ACM symposium on Principles of distributed
computing, pages 81--88, New York, NY, USA, 2007. ACM.

[6] M. Dinitz. My first 2.5 years of graduate school, 2008.

[7] J. M. Kleinberg, A. Slivkins, and T. Wexler. Triangulation and
embedding using small sets of beacons. In 45^th FOCS, 2004.

46

Track 5:

Software Engineering

Engineering Zombie Survival 47 Jones, L.A. World War C: The Rising
Threat of Undead Code.

Engineering Relentlessly 51 Beckman, Nels E. Relentless
Parallelism.

Engineering Mind Control 53 Leffert, Akiva. Provably Sound Orbital
Mind Control Lasers.

47

48

World War C: The Rising Threat of Undead Code

L.A. Jones

April 5, 2008

Abstract

Methods for detecting and eliminating "dead code" have been previously
dis cussed in the literature, but no attention has been given to the
increasing and far more problematic threat of "undead code." Undead
code spreads by con verting the surrounding "live" code into undead
code, thus spawning "zombie processes." To date, there have been
isolated incidents involving zombie pro cesses, the majority of which
were neutralized with relatively few casualties. However, a full-scale
outbreak of undead code can have serious consequences. Programs
infected with undead code consume memory and processor cycles as the
infection expands throughout the system, eventually devouring all
system resources. Left unchecked, an infestation of undead code will
turn its host into a "zombie computer," which will immediately begin
attacking other computers on the network in search of more processing
power. Undead code and zombie computers are extremely dangerous. This
paper presents the Headshot Method, an effective technique to
neutralize undead outbreaks that will aid researchers attempting to
control an onslaught of undead code.

49

50

ABSTRACT

Relentless Parallelism

Nels E. Beckman

Institute for Software Research

School of Computer Science

Carnegie Mellon University

nbeckman@cs.cmu.edu

sition. Other problems, unfortunately do not. These prob

It has become abundantly clear that, due to the rise of multi-core
architectures, parallelism is no longer a subject programmers can
ignore with impunity. Unfortunately, pro gramming concurrent code is
hard. I mean seriously. Some algorithms can not be parallelized, and
more importantly, some people cannot be bothered learning new
programming constructs. Toward returning to a state of programmer ig
norance, we present Relentless Parallelism, a programming methodology
that promises full utilization of all CPUs and cores without
additional programmer effort. We explain our system though an example
and formal rewriting rules.

1. INTRODUCTION

The field of computer science is currently in the midst of an all-out
crisis. Moore's Law, first formalized in 1965 contin ues to hold. The
number of transistors that can be placed on a process doubles
approximately every two years. However, we have reached the limit of
general-purpose performance for single CPU systems. Limiting factors,
for example heat, have made it increasingly difficult to utilize all
those new transistors in a single processor. Instead, ICU manufactur ers
have begun to develop multi-core CPUs, processors that internally
contain multiple distinct processors. Currently multi-core CPUs are
shipping with two and four cores, but the near future expects to see
dozens and even hundreds of cores per chip. Ladies and gentlemen, the
age of parallelism is upon us!

Unfortunately, the eminent scholars agree: Concurrency is Really, Really
Freaking Hard [1]. Developing applications that can actually take
advantage of many cores is poised to be the next great challenge of
computer science. In this paper we propose a programming methodology,
christened, Relentless Parallelism, that provides a solution to this
loom ing problem. Relentless Parallelism promises to keep each core in a
machine busy, even when developing algorithms for which no natural
parallel encoding exists.

This paper proceeds as follows: In Section 2 we explain relentless
parallelism by way of example of a traditionally hard-to-parallelize
algorithm, Huffman decoding. In Sec tion 3 we formalize this approach
using a series of rewriting rules. Finally, Section 4 concludes.

2. EXAMPLE: HUFFMAN DECODING Some algorithms, for example,
branch-and-bound search or optimization, lend themselves naturally to
parallel decompo

lems are particularly worrisome, since they will not be able to
benefit from the coming influx of CPU codes.

String huffmanDecodeByte(Queue\<Byte> byte_stream, DecTreeNode
cur_node) {

if( cur_node.getValue() != null ) {

// We are at a leaf node

return cur_node.getValue();

}

else {

if( byte_stream.remove().byteValue() == 0) { // Go to the left

return

huffmanDecodeByte(byte_stream,

cur_node.getLeftNode());

}

else {

// Go to the right

return

huffmanDecodeByte(byte_stream,

cur_node.getRightNode());

}

}

}

Figure 1: Huffman decoding: Because character codes have variable
lengths, a naive implementation is difficult to parallelize, for
example, using divide and-conquer.

Figure 2 is an example of one such algorithm, Huffman de coding.
Huffman coding is a prefix-free coding scheme, often used in
compression applications. In the scheme, characters are assigned
variable length codes based upon their probabil ity of appearance.
Since probabilities are allowed to change from case to case, a tree
mapping codes to characters is nec essary for decoding. The natural
way of decoding a series of bits is to proceed left or right down the
mapping tree (de pending on the current bit). When a leaf is reached,
that leaf necessarily specifies exactly one character, since the
scheme is prefix-free.

Unfortunately, because the length of codings is variable, par
allelizing this implementation is not straightforward. The normal
divide-and-conquer approach fails. If we were to di vide the bit
stream into multiple sections to give to multiple

51

cores, a seemingly natural fit, we would be unable to tell a priori
which size chunks to give to each processor, since one cannot tell
which bits denote the start or end of a character until decoding has
been performed.

Relentless Parallelism assures full utilization of each core even for
algorithms that are not naturally parallelize. Our technique consists of
a series of rewriting rules which add parallelism to otherwise
sequential algorithms. Figure 2 shows the result of this transformation
when applied to the Huffman decoding algorithm. Note that while Figure 2
shows the body of the huffmanDecodeByte, the result of the
transformation can only be seen at the top level of the pro gram.

String huffmanDecode(Queue\<Byte> byte_stream, DecTreeNode tree) {

class Parallelizer extends Thread {

public void run() {

for(int i=1, acc=1;

i\<this.hashCode();i++,acc*=1 ){}

this.run();

}};

int procs=

Runtime.getRuntime().availableProcessors(); for(int
i=0;i\<procs-1;i++) {

(new Parallelizer()).start();

}

StringBuffer result = new StringBuffer(\"\"); while(
!byte_stream.isEmpty() ) {

result.append(

huffmanDecodeByte(byte_stream, tree));

}

return result.toString();

}

Figure 2: The result of the Relentless Parallelism transform. Note how
the Parallelizer class pro duces maximum CPU utilization.

The result of the transform is that previously un-utilized CPUs are
now maximally utilized. The performance im provement is characterized
as follows:

Utilization_{0 =} 1

|CPUs|

Utilization_{rp =} |CPUs|

|CPUs|

3. FORMAL DESCRIPTION

In this section we provide formal rewriting rules for the Re
lentlessly Parallel programming system. These rules are de scribed in
Figure 3.

While the majority of the rules are relatively straight-forward, we
would like to draw special attention to the Asynch rule. We would expect
that our natural notion of parallelism would validate certain rules. One
of them is that channels can not

x and g do not alias

_{[x] := 1||[g] := 2} Concurrent Update

_{(f, q) : W → X} Morphism

[[(π)]]W ^[[p]]W

→ [[Q]]X

[[(π)]]X ^[[p]]X→ [[Q]]X

π P : Q, [[P]] : [[(π)]] ^.~→ [[Q]]~ Worlds

(h!0) \ h = δ when h /∈ P's channel

_{local h in (h!0; P) = P} Asynch

P(S × S)

P((S × S)^∞₎ Pom

Figure 3: The formal rewriting rules for the Relent less Parallelism
system.

affect the computation of processes that do not use them. This rule
shows that our notion of parallelism is correct.

4. CONCLUSION

The future of programming is an uncertain one. The rise of multi-core
architectures potentially will have vast and far reaching
consequences. A large majority of programmers are not familiar or
experienced writing parallel code. More over, some algorithms are not
easily parallelized, even by experienced coders. Yes it is a scary
future. However, in this paper we have presented a programming
methodology, Relentless Parallelism, that will help to remove much un
certainty from the future. Our methodology, which we have formalized
with a series of rewriting rules, will allow even sequential programs
to achieve maximum CPU utilization for all cores and processors.

4.1 Implementation

We have implemented this concept as a plug-in to the Eclipse Java
Development Tools IDE. This plug-in and source code are available for
download at the following address:

http://www.nelsbeckman.com/software.html

While the plug-in itself only works on Java code, rest as sured that
the monumental contributions we have made are applicable to any modern
programming language and For tran 77 [2].

5. REFERENCES

[1] Beckman, Nels E. Concurrency is Really, Really Freaking Hard. In
Proceedings of SIGBOVIK:

Workshop About Symposium on Robot Dance Party of Conference in
Celebration of Harry Q. Bovik's 0x40th Birthday. Pittsburgh, PA,
USA-A-OK. April 6, 2008.

[2] FORTRAN 77 4.0 Reference Manual. SunSoft
http://www.physics.ucdavis.edu/~vem/F77_Ref.pdf

52

Provably Sound Orbital Mind Control Lasers Akiva Leffert

Abstract {width="3.799332895888014in"
height="2.9076673228346457in"}

Human computation has been successful at tackling

problems that computers have had difficulty with.

However, this technique has many limitations. We

present a technique for easing or erasing these limi

tations and prove it sound.

1 Introduction

An increasingly popular technique for solving compu

tationally difficult problems is tricking humans into

doing it[7]. Some techniques, e.g. the ESPGAME[1]

frame these basically tedious tasks, in this case image

labeling, as games. This creates a reward for the user

in the form of a higher score. The Mechanical Turk[8]

pays humans for each small task performed. Finally,

RECAPTCHA[9] is used to protect web pages from

automated scripts while also performing valuable text recognition
activities.

The flaw in all of these techniques is that they require some sort of
reward structure. The user must enjoy the game. The user must need
money. The user must want to look at pornography. As a re sult, in order
to harness this computational power, we must have something of value.
Furthermore, this value must be higher than that of some other human
computation task from the perspective of the human. That means that all
human computation algorithms are subject to the whims of the populace.
Humans are notoriously fickle. It is hard to prove good bounds on human
behavior or get reliable uptime estimates.

A third flaw in these techniques is the limited re sources available for
human computations. People typically have jobs and families which
consume most of their cycles[3]. It is possible to construct more hu
mans, but the process is messy and inefficient. It is unclear whether
producing humans explicitly for the purpose of computation is
economically feasible [4].

In the remainder of this paper we present a tech nique for harnessing
human computation, the Orbital Mind Control Laser, and prove it sound.

Figure 1: A Hypnotoad

2 Related Work

There is a large body of work in the field of mind control. The area
is widely considered to have begun with a seminal series of papers by
Lamont Cranston in the 1930's under the general heading of Clouding
Men's Minds[6]. However, his technique only worked at a very small
scale. In the 1960's the CIA performed a variety of experiments in
this area but their results were unpredictable and their funding was
cut. The Illuminati actually perfected the science of mind con trol in
1975, but they do not publish[5]. It is a well known folk theorem
that sex can be used to control humans. The film, Invasion of the Bee
Girls doc uments an attempt to apply this to the small town scale.
Additionally, see Figure 1. The continued pop ularity of America's
Funniest Home Videos over the span of several decades also suggests
that some sort of mind control technique is being employed on a na
tional scale.

53

3 Results

{width="3.6851662292213474in"
height="3.6801662292213475in"}Figure 2: Relax

I count to three. One. Two. Three. Pafnuty.

In this section we present our mind control technique and its soundness
proof. Before continuing in this sec tion, relax. Clear your mind.
Breath deeply. Breath in. Breath out. Breath in. Breath out. Breath in.
Breath out. Breath in. Breath out. Breath in. Breath out. While
continuing to breath slowly and deeply, look at Figure 2 for one minute.
Do you feel calm? If not, look at the figure until you do. You're calm
now? Good. Now close your eyes and listen.

When you hear the word "Pafnuty"[2] all rational thought will vanish
from your mind. You will know only that this is a good paper; that it
should be ac cepted to your conference; that its author is brilliant,
dashing, and, dare I say it, sexy. You will want to give this paper good
reviews. You will write a good review and submit it to the program
committee. Then you will walk to your bank and send all of your money to
the following routing number: 3735928559. Do you have all of this?
Repeat it back to me. Very good. Furthermore, after you do this, you
will believe your self to be a platypus. Do you know how to behave like
a platypus? Okay, fine, just be a chicken. Keep breathing. You will know
and feel these things after

4 Conclusion

We presented a technique for harnessing the power of human
computation. Our proof was simple, clear, and revolutionary. You are
glad to have had a chance to read it. You love it. You love us.

References

[1] Unknown 1 and Unknown 2. Computer computa tion:how to detect
humans with tests that humans can generate and grade. Proceedings of
the 67^th Binarennial SIGBOVIK Conference, 2007.

[2] Pafnuty Chebyshev. � �� ��� ���� �������� ��� ���� �������,
1212.

[3] D. Gale and L. S. Shapley. College admissions and the stability
of marriage. American Mathematical Monthly, 69:9--14, 1962.

[4] T. Malthus. An Essay on the Principle of Popula tion, as it
affects the Future Improvement of Soci

54

ety, with Remarks on the Speculations of Mr God win, M. Condorcet and
Other Writers. J. John son, 1798.

[5] R. Shea and R. A. Wilson. Illuminatus! 1975.

[6] F. Street and F. Smith. That evil which lurks in the hearts of
men. 1931.

[7] L. von Ahn, M. Blum, and J. Langford. Telling humans and
computers apart automatically. 2004.

[8] Wolfgang von Kempelen.

[9] M. won Bhn and A. Spammer. Teaching spam mers to read.
Proceedings of the 67^th Binarennial SIGBOVIK Conference, 2007.

55

56

Track 6:

OMG Natural Language LOL

Restoring Language 59 McCann, James and Ronit Slyper. MADLIBS: The
MArkov reDacted Letter Interpretation B. System

Edday Anguageslay 63 Irshmanay, Ianbray Ray., Aurielay Away.
Onesjay, and Oesephjay Day. Onoughmcday. Igpay Atinlay inway Igpay
Atinlay: ethay igpay atinlay ictionaryday

orjectpay.

I HAS A LANGUAGE 75 Berenson, Dmitry. SOCIO-ECONOMIC FACTORS AND
AUTOMATED STATISTICAL ANALYSIS OF THE LOCAT LANGUAGE!!!1!

Wiki Language 79 Kua, John. General Case Rendering from Occurring
Instances.

57

58

MADLIBS: The **MA**rkov re**D**acted **L**etter **I**nterpretation
**B. S**ystem

James McCann^∗

Carnegie Mellon University Requires

Ronit Slyper^†

Carnegie Mellon University

Waxing Claudet

Liquid

Active

Snood Needs

My

_Destiny Garçon

Device

Abstract

Figure 1: Markov models encode the simple relationships between common
words.

introduction; §2, background information; §3, an algorithmic de

scription; §4, some results; and §5, the conclusion³.

We present a system to automatically guess redacted words in a censored
text by using context and domain knowledge. Our system uses a small
context around each removed word or phrase to build a model of the
word's contents. We find that our system is able to restore meaning to
many example corpi. ¹

CR Categories: M.I.6 [The Government Is Watching]: I Hear
Helicopters---Get Down!

Keywords: redacted, redtacted, redacted, redacted

1 Introduction

It is a well-documented fact that ever since the late 1950's, "the man"
[Tectonics -1e6] has been hiding things from us. Now, lately, it has
become popular to acquire snippets of "the man"'s [Leonard and King
1992] documents through Freedom of Information Act requests and routine
declassification. Of course the problem with these documents is that
"the man" [Inner Body 1999] has taken the trouble of removing certain
key words, phrases, and sen tences [Strunk and White 1999] from many
of these documents, for manly security reasons.

Thanks to the miracles of modern technology we can now, if not en tirely
restore these words, at least propose a maximum-likelyhood estimate of
their contents using a probabalistic inference model². In this paper
we present a simple model as well as some experimental results
demonstrating the efficacy of our approach.

This paper begins with an abstract, which is followed by: §1, the

^∗e-mail:jmccann@cs.cmu.edu

^†e-mail:rys@cs.cmu.edu

¹Or, for the non-CS literate: We present a cannibal to stuff about
redacted words in a murderous savage by using his socks there at. Our
system of boiling spout while around each other naked base kick to
mend that science of the word's contents. We find that our system is
able to stand no sofa of a native. Many thanks to Moby Dick for the
literary elevation.

²That is, we can guess.

2 Background

Probabilistic inference is a powerful technique for wrapping tech
nical verbage around blatent educated guessing. In vision, such a
framework bas been combined with the classical snake-balloon model
[Zhu et al. 1995].

3 Algorithm

Our algorithm proceeds in two phases, which we term adolescence and
out of. In the adolescent phase, we build a frequency count table
for co-occuring words. These words are drawn from domain specific
sample texts. In the out of phase, the censored text is pre processed
to assess the number of words redacted in each segment. Finally,
posthumously, the contents of each redacted segment is ex tracted by
dynamic programming.

Our system is implemented in Perl. We plan on releasing the source as
soon as our visas to Xanth come through.

3.1 Adolescence

We build our model by training our system on a corpus [Musil and
Mirsky 1914] of text. During this training phase we perform fre
quency counts of the occurrance of words. These counts are stored in a
hash table [Glenda 2001].

3.2 Out of

A shortest-path algorithm on log-likelyhood is used to fill the con
text, with randomization breaking ties to reality.

³While such summary sentences hold no actual content, they do take
up valuable column inches.

59

Figure 2: Clearly linear.

{width="2.0001662292213473in"
height="1.501000656167979in"}

Table 1: These results chaired us up immensely.

3.3 Rigorous Evaluation

---

---

Despite ing IRB approval, we performed a rigor

---

---

ous user-study with consenting users. Results

---

---

---

---

were , as ex pected (see Figure 2 and Table 1).

4 Results

We present several example redacted texts [CIA 1971; Silverstein
1970], as seen in Figure 4, Figure 6, and Figure 5. These show the
method is strong enough to have practical applications, such as in the
"My dog redacted my homework", "The NSA redacted my resume", and the
increasingly-common "My university redacted my tuition bill" situations.

5 Conclusions

We have demonstrated a method of removing most of the ambiguity from a
wholly redaction-filled document. Our method depends on having an
appropriate corpus. ⁴

⁴Or, We have bejuggled a method of removing the unknown stranger
captain from a wholly redaction-filled document. Our method depends on
having an irregular cursings.

+-----------------------------------------------------------------------+
| > In A.D. 2101, war was beginning. |
| > |
| > What repair if ?!? |
| > |
| > Somebody set and lusty days to store thou get signal. What ! |
| > |
| > Mai[n screen turn on.]{.underline} |
| > |
| > It's and bristly beard then . |
| > |
| > How are from thy gentlemen !! |
| > |
| > All your base are the world us. |
| > |
| > You are from that on to destruction. |
| > |
| > What you should that which ? |
| > |
| > You have no chance to survive make confounds in . |
| > |
| > Ha end and Ha .... |
| > |
| > Cap_~tain!!~ |
| > |
| > Take every where every 'ZIG' !! |
| > |
| > You know all the grave . |
| > |
| > Move 'ZIG['.]{.underline} |
| > |
| > For great with . |
+=======================================================================+
+-----------------------------------------------------------------------+

Figure 3: Uncensoring All your base are belong us using Shake
speare's Sonnets.

Acknowledgments

This work was supported by an oppresive sense of paranoia, and a
repressive and censorious political climate.

References

CIA, 1971. Family jewels. http://www.gwu.edu/
nsarchiv/NSAEBB/NSAEBB222/index htm.

GLENDA, M. 2001. Eat All You Want No Weight Gain Breakfast Cookbook.
Hash Browns 3 Ways.

INNER BODY, 1999. The male. http://www.innerbody.com/image/repmov htm.

LEONARD, B., AND KING, S., 1992. The lawnmower man. "God made him
simple. Science made him a god.".

MUSIL, R., AND MIRSKY, M. 1914. Diaries. Entry: 11 June Medics'
jargon.

SILVERSTEIN, S., 1970. The bagpipe who didn't say no. STRUNK, W., AND
WHITE, E. B. 1999. The Elements of Style. TECTONICS, P., -1e6. Ellan
vannin. 54^◦09' N, 4^◦29' W.

ZHU, S., LEE, T., AND YUILLE, A., 1995. Region competition: Unifying
snakes, region growing, energy /bayes/mdl for multi band image
segmentation.

60

+-----------------------------------------------------------------------+
| > The Bagpipe Who Didn't Say No. |
| > |
| > It was nine o'clock at midnight at a quarter after three |
| > |
| > When a turtle met so in that was bound by the sea, |
| > |
| > And to many said, "My dearie, |
| > |
| > May I sit with you? I'm stronger And the violence didn't surges |
| > have Said the turtle to you and tomorrow I have walked this lonely |
| > shore, |
| > |
| > I have talked to waves and pebbles--but I've never the chance Will |
| > you marry me today, dear? |
| > |
| > Is it 'No' you're going to say dear?" |
| > |
| > But like embryonic didn't say no. |
| > |
| > Said the turtle to iraq have no Please excuse me if I stare, |
| > |
| > But you have leaders and hold dear, |
| > |
| > And you have the strangest ahead If I begged people our whole |
| > nation Could I give you just one squeeze, love?" And a nation |
| > didn't say no. |
| > |
| > Said the turtle and eventually reverse the Ah, you love me. Then |
| > confess! |
| > |
| > Let me whisper in your dainty ear and you and reform our prosperity |
| > And he cuddled enemies agree on her And so lovingly he squeezed |
| > her. |
| > |
| > And to many said, " have Said the turtle to have the enemy Did you |
| > honk or bray or neigh? |
| > |
| > For 'Aaooga' when your kissed is such a heartless thing to say. |
| > |
| > Is it that I have offended? |
| > |
| > Is it t[hat our love is e]{.underline}nded?" |
| > |
| > And your freedom didn't say no. |
| > |
| > Said they have to a year our Shall i leave you, darling wife? |
| > |
| > Shall i waddle off iraqi surges Shall i crawl out of your life? |
| > |
| > Shall I move, depart and go, dear-- |
| > |
| > Oh, I beg you tell me 'No' dear!" |
| > |
| > But in to didn't say no. |
| > |
| > So the turtle crept off crying and he ne'er came back no more, |
| > |
| > And he left saw our lying on that smooth and sandy shore. |
| > |
| > And some night when you is by progress Just walk up and say, |
| > "Hello, there," |
| > |
| > And politely ask the time if this story's really so. |
| > |
| > I assure you, darling children, include foreign won't say "No." |
+=======================================================================+
+-----------------------------------------------------------------------+

Figure 4: Uncensoring The bagpipe didn't say no. Boxed text was
redacted and filled in by our system. The method was trained on the
2007 state of the union address.

+-----------------------------------------------------------------------+
| > Lastly, before I sign off, our diplomats feText of all you by U.S. |
| > Official in Iraq Posted ar using leverage. It is much nicer to |
| > sleep at the resort gave a very clear appropriated for his own |
| > personal use when you don't have to listen to him harp and |
| > complain. Likewise, it is better to keep we are what a happy drunk |
| > rather than an angry drunk. If our diplomats and CPA officials feel |
| > uncomfortable being bad cop, it is essential that people in |
| > Washington play the role. lifted himself and |
| > |
| > you are for example, are much more compliant when their checks are |
| > "delayed" or fail to appear. The same is true with other Governing |
| > Council members. The key is subtlety. They will figure out the |
| > connection on their own; they need not have it pointed out by |
| > Bremer or Greenstock in a way that will cause them to dig in their |
| > heels. |
+=======================================================================+
+-----------------------------------------------------------------------+

Figure 5: Portion of a memo on Iraq, as unredacted with frequency
counts of various Dr. Seuss texts.

61

+-----------------------------------------------------------------------+
| > MEMORANDUM FOR THE RECORD |
| > |
| > SUBJECT: milk Equipment Test, Miami, Florida, August 1971 |
| |
| The following details concerning the wagons arrangements for Subject |
| tests |
| |
| > were provided by the thneeds and then during a telephone |
| > conversation with the undersigned, 7 May 1973. |
| > |
| > look lorax now retired, formerly assigned to the grass was the |
| > trees for the August 1971 Field Test of the chopping as it from |
| > Security arrangements for the test were handled on behalf of |
| > turtles and the you visitors by the throne in conjunction with the |
| > trees Security Officer, who was just a tree at the time. that |
| > |
| > was in daily contact with let them Miami Police in the course of |
| > his official liaison duties. |
| > |
| > ler family was reluctant to call i just at home over an open |
| > telephone line to inquire about the specifics of the a arrangements |
| > at this point, and suggested that the so Security Officer by this |
| > time might have been transferred back to Headquar_~ters~ and be |
| > available for a direct query. |
| > |
| > The writer called care give DIV/D Security officer, who verified |
| > the fact that all happy indeed is stationed at Headquarters, with a |
| > current assignment to a |
| > |
| > king lifted the located in the lifted lorax and on is available via |
| > the following telephone connections: |
| > |
| > - and sour when |
| > |
| > The above details were provided by telephone to lifted his |
| > gruvvulous Chief, Division D at 1650 hours this date. |
| > |
| > (signed) ought to . |
+=======================================================================+
+-----------------------------------------------------------------------+

Figure 6: A CIA memo, uncensored using Dr. Seuss texts.

62

Igpay Atinlay inway Igpay Atinlay:

ethay igpay atinlay ictionaryday orjectpay

Ianbray Ray. Irshmanhay

Epartmentday ofway Omputercay Iencescay

Arnegiecay Ellonmay Universityway

Ittsburghpay, Apay 15213

irshmanhay@cs.cmu.edu

Aurielay Away. Onesjay

Epartmentday ofway Omputercay Iencescay

Arnegiecay Ellonmay Universityway

Ittsburghpay, Apay 15213

onesjay@cs.cmu.edu

Oesephjay Day. Onoughmcday

Epartmentday ofway Assicsclay

Ethay Entkay Oolschay

Entkay, Tcay 06757

onoughmcjday@kent-school.edu

Abstract

Igpay Atinlay isway away inefay andway eputableray ialectday ofway
ethay Englishway anguagelay. Oughthay ethay ialogday ashay eenbay
aroundway orfay undredshay ofway earsyay, ethay authorsway ofway
isthay aperpay avehay otnay eenbay ableway otay indfay away igpay
atinlay - igpay atinlay ictionaryday. Eway, otway udentsstay ofway om
putercay iencescay andway oneway esteemedway eachertay ofway atin lay,
avehay oticednay isthay istoricalhay oversightway andway avehay
ecidedday otay ectifyray ethay ituationsay. Eway erehay escribeday our
way indingsfay omfray ourway ictionaryday ompilationcay ocesspray
andway esentpray ourway esultsray orfay useway ybay indergardener
skay, elementryway oolersschay, andway ighhay-oweredpay esearcher sray
orldway-ideway.

Igpay Atinlay [play] isway away ommoncay ildrenchay'say amegay
ayedplay inway ass roomsclay, arkspay, oolyardsschay, andway
aygroundsplay acrossway ethay Englishway eakingspay orldway [1].
Oughthay isthay ackbay-angslay anguagelay appealsway imari lypray otay
ethay oungeryay enerationgay ofway Englishway eakersspay, itway ashay
an way appropriateway audienceway amongway osethay overway
entytway-ivefay, including way oolschay eacherstay, andway arentspay
ofway oungyay ildrenchay, andway inguisti clay esearchersray [2].
Orfay osethay owhay oday avehay otnay eenbay exposedway otay play,
ethay authorsway uggestsay atthay ethay eaderray eakspay ibberishgay
otay anyway

Englishway-eakingspay eightway-earyay-oldway; ifway ethay atternpay
ofway ibberish gay epeatedray ackbay isway ubjectsay otay away
egularray atternpay ofway eechspay, itway ethay authorsway'
ollectivecay uspicionsay atthay ethay esponseray isway inway Igpay
Atinlay ithway obabilitypray asymptoticallyway approachingway unityway
asway ethay engthlay ofway ethay ildchay'say esponseray increasesway.
Ilewhay empiricalway oofpray ofway isthay uppositionsay annotcay ebay
independentlyway erifiedvay inway ethay iteraturelay, itway isway
ellway ownknay atthay play isway away ommoncay ialect day.

Oughthay ethay play anguagelay ashay eenbay aroundway orfay away
eryvay onglay imetay, ethay authorsway -- inway onductingcay away
iteraturelay eviewray -- ealizedray atthay away Igpay Atinlay
ictionaryday ashay otnay eenbay ittenwray. Ilewhay erethay areway
umerousnay Englishway otay Igpay Atinlay anslatorstray [3, 4],
omesay ofway ichwhay areway urveyedsay inway ethay ackgroundbay
ectionsay, ethay authorsway eal izedray atthay away uetray
play-inway-play ictionaryday oesday otnay existway. Onsid eringcay
ethay idespreadway useway ofway Igpay Atinlay amongway ethay oungeryay
enerationgay, away enerationgay inway ichwhay individualsway areway
oremay ikelylay anthay averageway otay absorbway ictionariesday inway
away agepay-byay-agepay ash ionfay, away elfsay-efinedday
exicographiclay agglomerationway ofway Igpay Atinlay er minologytay
isway earlyclay overdueway. Additionallyway, incesay otherway
udiesstay avehay ownshay atthay ictionaryday useway eatlygray
increasesway ildrenchay'say ead ingray andway ellingspay abilityway,
away play ictionaryday ouldshay ebay elcomedway asway away aluablevay
educationway ooltay [5].

Isthay ojectpray oughtsay otay onstructcay uchsay away Igpay Atinlay
ictionaryday, aking tay advantageway ofway ethay omputationalcay
esourcesray availableway otay ethay ad uategray udentsstay atway
Arnegiecay Ellonmay ilewhay imultaneouslysay everaginglay ethay
assicsclay expertiseway ofway eirthay ollaboratorcay atway ethay
Entkay oolschay. Inway osay oingday, itway eekssay otay ingbray
orderway otay away anguagelay atthay asway eviouslypray asedbay
olelysay onway ethay uancesnay ofway Englishway, andway aymay avehay
ethay (erhapspay unfortunateway orfay adultsway) onsequencecay ofway
eatingcray away anguagelay ofway, byay, andway orfay ildrenchay [2].

Ethay emainderray ofway isthay aperpay isway organizedway asway
ollowsfay. Ection say 1 iscussesday astpay orkway onway Igpay Atinlay,
includingway umerousnay astpay attemptsway atway anslationtray
ictionariesday atthay allfay ortshay ofway ourway ethay Igpay
Atinlay - Igpay Atinlay andardstay. Ectionsay 2 escribesday ourway
ocedurepray or fay ompilingcay ethay Igpay Atinlay ictionaryday.
Ectionsay 3 escribesday ourway esult sray, includingway aggregateway
atisticsstay ofway ourway ictionaryday. Ectionsay 4 on cludescay.

1 Ackroundbay

Erethay areway everalsay ellway-ecognizedray ialectsday ofway Igpay
Atinlay. Accord ingway otay Ikipediaway, ethay orldway'say ostmay
eliableray ourcesay orfay informa tionway, ethay ajormay ariantvay
ofway Igpay Atinlay anslatestray anway Englishway or dway intoway
Igpay Atinlay byay emovingray ethay initialway onsonantcay orway onso
nantcay usterclay otay ethay endway ofway ethay ordway andway
appendsway away away onglay-Away oundsay [?, 6]. Orfay osethay otnay
amiliarfay ithway Igpay Atinlay onun ciationpray, isthay isway
onouncedpray eIway accordingway otay ethay Internationalway Oneticphay
Alphabetway (IPAway) [7]. Oughthay Igpay Atinlay isway oftenway away
okenspay anguagelay, Igpay Atinlay isway anscribedtray usingway ethay
andardstay Atin lay alphabetway andway itingwray ethay onglay-Away
asway "ayway," oughthay IPAway uristspay aymay aimclay atthay uchsay
away anscriptiontray isway ambiguousway [7]. Ifway away ordway
oesday otnay avehay anway initialway onsonantcay, ethay andardstay

Englishway ordway isway onouncedpray, andway enthay away "ayway"
oundsay (IPAway: eIway) isway appendedway [7]. Otablenay ariantsvay
ofway Igpay Atinlay useway away "ayhay" oundsay (IPAway: eIhay) orway
"ayyay" oundsay (IPAway: eIjay). Otherway ari antsvay enatehyphay
ethay ailingtray onsonantcay usterclay orway anslatingtray otherway
oundssay uchsay asway "eshway" intoway Igpay Atinlay ariantsvay [7].

Igpay Atinlay onunciationpray, oweverhay, isway ownknay otay aryvay
ithway eography gay, away eaturefay atthay akesmay itway anway
interestingway orfay inguisticslay udys tay [8, 1]. Inguisticslay
orkway ashay imarilypray ocusedfay onway ethay owelvay iftsshay inway
Igpay Atinlay, ebasingday ethay allowedhay andway oftylay anguagelay
inway ope shay ofway uncoveringway inormay ubtletiessay ofway ethay
Englishway onguetay. Uch say inguisticlay ilferingpay isway, alasway,
aracteristicchay ofway Englishway [2] -- and way Englishway
inguistslay, itway eemssay, avehay eenbay ethay ostmay egregiousway
ofway allway. Everthelessnay, itway isway unfairway otay ismissday
uchsay orkway out way ofway andhay. Inguisticlay orkway onway Igpay
Atinlay ashay uggestedsay atthay erethay ereway ifferentialday
atterningpay orfay onsonantcay equencessay inway Igpay Atinlay, orway
ashay arguedway orfay ethay utilityway ofway ethay ecedencepray
odelmay ofway onologicalphay epresentationsray [8, 1]. Etherwhay
orway otnay ethay owledgek nay ainedgay isway orthway ethay
inguisticlay amageday isway, erhapspay, away ild chay'say ebateday.

Oughthay eatedtray oughlyray byay ethay inguistslay, play ashay eenbay
eceivedray ore may indlykay byay otherway anchesbray ofway ethay
academyway. Athematicsmay, orfay instanceway, ashay akentay entlegay
otenay ofway Igpay Atinlay. Orfay instanceway, athe maticiansmay
avehay owshay play otnay ebay oneway-otay-oneway: ethay ordsway orfay
omethingsay atthay oneway atchesscray asway ellway asway away omanway
owhay ies flay aroundway onway away oomstickbray othbay anslatetray
otay "itchway", atway east lay inway ethay ostmay ommoncay Igpay
Atinlay ariantvay [7]. Ilewhay omesay olarss chay avehay entspay
ubstantialsay effortway ingtryay otay owshay atthay Igpay Atinlay
isway omplexcay, irrationalway artiallypay orderedway, orway
ellway-oundedfay, uchsay oblemspray emainray openway questionsway
andway avehay ontinuedcay otay efuddlebay olarsschay andway avehay
emainedray openway esearchray questionsway [2]. Orfay oth erway
oofspray inway play, oweverhay, ethay eaderray isway advisedway otay
openway anyway athematicsmay extbooktay usedway inway away
ollegiatecay orway aduategray evellay oursecay [2].

Ivengay ethay extensiveway useway ofway Igpay Atinlay inway
athematicsmay, itway isway onay urprisesay atthay omputercay
ientistsscay avehay egunbay otay ollowfay uit say. Igpay Atinlay ashay
eceivedray attentionway omfray omputercay iencescay esearcher sray
orfay uchmay ofway ethay isciplineday'say ifetimelay. Igpay Atinlay
ashay een bay advocatedway asway anway importantway ooltay orfay
eachingtay introductoryway ogrammingpray, artiallypay ecausebay ethay
egularray ucturestray ofway ethay anguage lay isvay away isvay ethay
Englishway anguagelay akesmay itway away imepray exam pleway orfay
eachingtay ecursionray [7]. Orfay instanceway, Igpay Atinlay ashay
eenbay usedway asway away implesay eachingtay exerciseway orfay Isplay
ogrammingpray [9], oughthay omesay objectway-orientedway
ogrammerspray avehay iedtray otay udgeonblay eirthay ayway oughthray
play usingway Avajay. play anslationtray ashay eenbay away allmarkhay
ofway introductoryway omputercay iencescay assesclay atway umerousnay
ol legecay ampusescay orldway-ideway, osay uchmay osay atthay
"ellohayway orldwayway" aymay, erhapspay, ivalray ethay ACMway'say
"ellohay orldway" ojectpray [10]. Igpay Atinlay ashay alsoway eenbay
ofway interestway otay eechspay esissynthay esearchersray ueday otay
ethay egularray ucturestray ofway ethay astlay ablesyllay [5]. Orfay
instance way, esearchersray atway Ellbay Abslay entspay away
ubstantialsay amountway ofway imetay evelopingday away
exttay-otay-eechspay emsystay orfay play, oughthay ethay ur rentcay
implementationway onway eirthay ublicpay itesay appearsway otay ebay
onnay-

unctionalfay asway ofway isthay intingpray [5].

Oughthay omputercay ientistsscay avehay ometimessay eenbay accusedway
ofway eing bay interestedway inway oblemspray atthay avehay onay
impactway onway ethay ealray orldway [2], isthay ashay otnay eenbay
ethay asecay ithway csay interestway inway play. Umerousnay Igpay
Atinlay anslatorstray avehay eenbay evelopedday andway areway avail
ableway orfay eefray onlineway [4, 3, 9]. Ooglegay'say Igpay Atinlay
earchsay engineway allowsway orfay ebway earchessay inway Igpay
Atinlay, away eryvay usefulway ooltay orfay ethay ommunitycay.
Indeedway, Ooglegay allowsway away userway otay accessway allway
itsway agespay inway Igpay Atinlay; oweverhay, auxiliaryway ervicessay
uchsay asway Ooglegay Ailmay avehay etyay otay ebay updatedway otay
aketay advantageway ofway ethay anguagelay [4]. Igpay Atinlay
anslatingtray oftwaresay isway availableway orfay eefray atway away
arietyvay ofway acesplay, includingway Ooglegay [3, 4]. It way ashay
eenbay uggestedsay atthay ethay ostmay uccessfulsay achinemay
anslationtray ojectpray otay ateday ashay eenbay inway ethay
Englishway otay Igpay Atinlay omainday; espiteday ecentray ogresspray
ademay inway anslatingtray Englishway otay otherway an guageslay,
evenway optay esearchersray oncedecay atthay Igpay Atinlay eakersspay
illway ontinuecay otay emainray inway eirthay uniqueway andway
enviableway ositionpay orfay away ubstantialsay amountway ofway
imetay.

Espiteday ethay eatgray ainsgay ademay byay omputercay ientistsscay
inway ethay Igpay Atinlay omainday, away ubstantialsay amountway ofway
orkway emainsray undoneway. Away iteraturelay urveysay uggestedsay
atthay erethay ashay otnay eenbay away oncert edcay effortway otay
utpay ogethertay away Igpay Atinlay ictionaryday. Ilewhay ethay
ideaway ofway ethay ictionaryday itselfway isway ignificantlysay
olderway anthay ethay Igpay Atinlay anguagelay, usthay arfay erethay
ashay otnay eenbay ignificantsay orkway otay associateway
efinitionsday ithway ethay ordsway okenspay inway isthay anguage lay.
Englishway-Englishway Ictionariesday ateday ackbay otay atway eastlay
ethay even teenthsay enturycay, oughthay onesway associatingway
Englishway ordsway ithway Atin lay onesway avehay existedway orfay
away uchmay ongerlay eriodpay ofway imetay [11]. Ethay acklay ofway
away Igpay Atinlay Amuelsay Ohnsonjay ashay eftlay ethay anguage lay
inway away alaisemay, andway itway isway isthay oblempray atthay eway
eeksay otay addressway inway isthay orkway.

2 Ethodsmay

Ethay ictionaryday eatedcray inway isthay ojectpray asway eatedcray
omfray ethay On lineway Ainplay Exttay Englishway Ictionaryday
(OPTEDway) ublicpay omainday ic tionaryday ersionvay 0.003. Ethay
ictionaryday isway asedbay offway ofway ersionvay 0.47 ofway ethay
Ollaborativecay Internationalway Ictionaryday ofway Englishway, ich
whay asway inway urntay erivedday omfray ethay 1913 ersionvay ofway
Ebsterway'say Evisedray Unabridgedway Ictionaryday andway
upplementedsay ithway additionalway or dway efinitionsday omfray
OrdNetway [12]. Ethay ictionaryday asway eleasedray byay Alphray
Utherlandsay andway isway underway ethay Ugnay ublicpay icenselay
[13]. Ile whay ethay authorsway ecognizeray atthay isthay
ictionaryday ontainscay ermstay atthay areway atedday, andway omesay
ofway ethay ientificscay efinitionsday avehay eenbay ub sequentlysay
ownshay otay ebay inaccurateway orway ongwray, ethay exttay usedway
inway isthay ojectpray epresentsray ethay ostmay ompletecay
ictionaryday atthay isway ubliclypay availableway inway anway
easilyway arsiblepay ormatfay.

Omesay inormay angeschay adhay otay ebay ademay otay ethay inputway
ilesfay, includ ingway ethay emovalray ofway away umbernay ofway
aracterschay atthay ouldcay otnay ebay arsedpay byay ethay ASCIIway
arserpay; oweverhay, ethay umbernay ofway ordsway odifiedmay asway
esslay anthay away undredhay. Urthermorefay, omesay ordsway ere way
incorrectlyway enatedhyphay, esultingray inway emthay appearingway
inway outway-

ofway-orderway ositionspay inway ethay ictionaryday; esethay ordsway
ereway odified may eforebay ethay inalfay iptscray asway unray.
Astlylay, asway inway allway oodgay ictionariesday, away ecialspay
ordway asway addedway otay ethay ictionaryday inway or derway otay
otectpray againstway opyingcay andway agiarismplay [14]. Indingfay
esethay ordsway areway eftlay asway exercisesway otay ethay eryvay,
eryvay uriouscay eaderray.

Ethay Englishway ictionaryday asway appedmay otay Igpay Atinlay
usingway away Erlpay iptscray. Ethay iptscray asway ittenwray usingway
Erlpay 5.10 andway onsistedcay ofway aboutway otway undredhay ineslay
ofway odecay. Inway onformancecay ithway ethay Ug nay ublicpay
icenselay, ethay ourcesay odecay orfay ictionaryday eationcray isway
avail ableway omfray ethay authorsway. Ethay Erlpay iptscray asway
unray onway away uster clay ofway erverssay elongingbay otay ethay
oolschay ofway omputercay iencescay atway Arnegiecay Ellonmay
Universityway. Orfay oodgay easuremay, ethay ictionaryday on
structioncay ocesspray utilizedway onay esslay anthay ivefay erverssay
ereway usedway iavay emoteray esktopday, oughthay itway isway
arguableway atthay away oderatelymay owerfulpay aptoplay ithway
aboutway 50MB ofway ardhay iskday acespay ouldway ave hay ufficedsay.
Incesay ildrenchay areway onepray otay quotingway atisticsstay, andway
incesay entay-earyay-oldway play eakersspay areway especiallyway
onepray otay eliev ingbay atthay oremay isway alwaysway etterbay,
ethay authorsway osechay otay artifi ciallyway inflateway eirthay
atisticsstay incesay ethay esourcesray ereway availableway.
Omputationcay ooktay esslay anthay away inutemay.

3 Esultsray

Ethay ictionaryday eatedcray isway away otaltay ofway ivefay
ousandthay, ourfay un dredhay, andway elvetway agespay onglay andway
equiresray 21,595kB ofway iskday acespay otay orestay. Itway
ontainscay onway ethay orderway ofway 180,000 ordsway andway 2.0
illionmay ordsway orthway ofway (osepray) efinitionsday inway ethay
ocu mentday. Unlikeway Englishway ictionariesday, ichwhay avehay
ordsway ichwhay artstay ithway anyway oneway ofway entytway-ixsay
etterslay, ethay Igpay Atinlay ictionaryday ashay ordsway atthay
onlyway eginbay ithway Away, Eway, Iway, Oway, orway Uway. Orfay ethay
enefitbay ofway Englishway eakersspay, oweverhay, ethay ordsway ereway
organizedway uchsay atthay entriesway ouldcay ebay oundfay usingway
ethay English way ellingsspay. Oughthay Igpay Atinlay uristspay avehay
oicedvay eirthay omplaintscay aboutway isthay estylay ofway
organizationway, itway asway uggestedsay atthay uchsay anway
organizationway ouldway ebay easiestway orfay onnay Igpay Atinlay
eakersspay otay useway. Away uturefay ictionaryday, esignedday orfay
advancedway orway uentflay Igpay Atinlay eakersspay, isway urrentlycay
inway evelopmentday.

Otway amplesay ictionaryday agespay areway attachedway asway iguresfay
1 andway 2. Espiteday iticcray aimsclay otay ethay ontrarycay, esethay
agespay ereway otnay insert edway intoway isthay ocumentday olelysay
otay ovidepray engthlay. Ethay authorsway ishway otay ointpay outway
ethay educationalway, idacticday, andway inguisticlay alue vay ofway
insertingway ictionaryday agespay intoway academicway ocumentsday.

Away umbernay ofway importantway ictionaryday actsfay areway inway
orderway. Ethay irstfay ordway ofway ethay ictionaryday isway away,
andway ethay astlay ordway isway umzythay. Ethay efinitionday ofway
ostmay ordsway itsfay onway oneway inelay only way, oughthay erethay
areway everalsay ordsway (includingway ethay irstfay ordway in way
ethay ictionaryday) atthay eednay inenay ineslay orfay ethay
efinitionday. Ifway away ordway ashay ultiplemay efinitionsday,
eachway efinitionday isway includedway onway away eparatesay inelay.
Ostmay ordsway onlyway avehay oneway efinitionday, oughthay omesay
ordsway (ikelay ethay irstfay ordway) avehay inenay orway oremay.
Ethay ord way otistpray ancay ebay oundfay onway agepay 3726. Ethay
astlay ordway onway age pay 763 isway entercay. Ethay ordway undredhay
oesday otnay appearway onway agepay

{width="4.724332895888014in"
height="6.247666229221347in"}Figure 1: Away andomray ictionaryday
agepay

{width="4.724332895888014in"
height="6.247666229221347in"}

Figure 2: Anotherway pageway. Ancay ouyay insertway ictionaryday
agespay inway our way academicway aperspay?

oneway undredhay, utbay atherray onway agepay 2268. Ethay asephray
Igpay Atinlay asway otnay includedway inway ethay ictionaryday
incesay, ifway ouyay eadray ethay ic tionaryday, ouyay'llay earnlay
atwhay itway isway allway ightray. whyay anyway ofway esethay
idbitstay areway importantway areway otnay articularlypay elevantray;
owever hay, ownay atthay eythay areway ointedpay outway, ethay avidway
ictionaryday eaderray illway ebay oremay ikelylay otay ooklay emthay
upway. Additionallyway, byay ointingpay esethay outway, ethay avidway
eaderray illway ebay oremay ikelylay otay owsebray otay ethay
ictionaryday agespay aroundway esethay ordsway andway artstay
ookinglay orfay istakesmay. Ereforethay, ethay authorsway illway otnay
ebay urprisedsay ifway eythay eceiveray emailsway omfray idskay ithway
ayway otay uchmay imetay onway eirthay andshay (owhay elseway) owhay
avehay othingnay etterbay otay oday anthay ickpay on way academicsway
inway eirthay ivoryway owerstay byay ointingpay outway erewhay
eythay'evay onegay ongwray. Ethay authorsway avehay eensay itway
eforebay, idskay, andway unlikeway Uthknay eythay arenway'tay oinggay
otay ivegay ouyay away ickelnay erpay istakemay ouyay indfay.

Onay aperpay ouldway ebay ompletecay ithoutway enchmarkingbay
atisticsstay, andway isthay aperpay intendsway otay ebay onay
ifferentday. Irshmanhay'say Unabridgedway Ig pay Atinlay Ictionaryday
asway omparedcay otay oneway authorway'say ersonalpay opy cay ofway
Ebsterway'say Inthnay Ewnay Ollegiatecay Ictionaryday, intagevay 1983
[15]. Ilewhay ethay authorsway idday otnay otherbay otay ountcay
ethay ordsway inway Ebster way'say ictionaryday (ethay editorsway
aimedclay atthay eythay adhay irteenthay unique way illionmay
ordsway), eythay idday oticenay atthay Ebsterway'say ictionaryday
adhay onlyway 1562 agespay. Ebsterway'say ictionaryday alsoway
includedway icturespay, aw ingsdray, andway otherway icturespay; itway
asway alsoway otednay atthay Ebsterway odifiedmay ethay arginsmay,
angedchay ethay ontsfay, andway ayedplay otherway eset tingtypay
amesgay. Ethay authorsway ofway ethay Igpay Atinlay ictionaryday,
ereforethay, aimclay atthay eirthay ictionaryday isway ethay
academicallyway uperiorsay ocumentday -- andway ereforethay, ouldshay
eceiveray ethay igherhay adegray -- incesay eythay idday otnay
esortray otay oolboyschay amesgay.

Ethay efacepray agepay ofway ethay ictionaryday ontainscay away Igpay
Atinlay onnet say. Isthay isway otnay away eaturefay oundfay inway
otherway ictionariesday, andway ereforethay akesmay Irshmanhay'say
Unabridgedway Igpay Atinlay Ictionaryday early clay uperiorsay otay
anyway otherway ictionaryday -- Igpay Atinlay orway otherwiseway --
atthay isway availableway odaytay.

4 Onclusioncay

Ethay authorsway ereway ableway otay uccessfullysay ableway otay
eatecray eirthay Igpay Atinlay ictionaryday, ethay irstfay
Igpay-Atinlay-inway-Igpay-Atinlay ictionaryday etyay ownknay. Inway
osay oingday, eythay avehay illedfay away apgay inway ethay Igpay
Atinlay iteraturelay, ovidingpray away ecessarynay ervicesay orfay
play eakersspay orld wideway.

Away eviewerray ommentedcay atthay ethay ellingspay inway isthay
ocumentday aymay ebay uspectsay, especiallyway ivengay ethay acklay
ofway appropriateway ellspay ecker schay orfay ethay anguagelay.
Ilewhay ethay authorsway acknowledgeway andway ake tay
esponsibilityray orfay anyway ostypay inway ethay ocumentday, inway
ethay ourse cay ofway eirthay orkway ethay authorsway ereway ableway
otay indfay away umbernay ofway imitationslay inway ethay Igpay
Atinlay ammargray emsystay. Irstfay, play ackslay away uccinctsay
andway ellway-establishedway uleray orfay ealingday ithway ossessive
pay asescay. Ethay authorsway avehay osenchay ethay andardstay ofway
acingplay ethay "apostropheway-say" iorpray otay ethay erminaltay
"ayway", esultingray inway expres sionsway uchsay asway "ethayway
authorsway'ayway eliefsbayway", althoughway eythay

ereway unableway otay indfay away ecedentpray orfay isthay inway ethay
extensiveway Igpay Atinlay iteraturelay. Imilarlysay, play alsoway
ackslay away etsay ofway ulesray ecificallyspay ealingday ithway
enatedhyphay ulesray itway asway ecidedday atthay allway ortionspay
ofway ethay enatedhyphay ordway ouldway ebay enatedhyphay otay
eatecray asesphray uchsay asway "errymay-ogay-oundray". Astlylay,
ethay authorsway ereway un ableway otay indfay away etsay ofway
ulesray orfay ealingday ithway onnay-Englishway ordsway inway Igpay
Atinlay. Ethay authorsway' esponseray otay isthay asway otay applyway
away Igpay-Atinlay entriccay approachway otay oreignfay ordsway,
evenway oughthay erethay areway otherway ialectsday (uchsay asway
oucherbemlay orfay Ench fray orway attenenglischmay inway Ermangay)
ichwhay ightmay avehay eenbay oremay appropriateway.

Ethay orkway oneday erehay isway onlyway away allsmay epstay [2],
andway evenway omethingsay asway onumentalmay asway away ictionaryday
isway utbay away allsmay epstay inway ethay astvay ainchay ofway
umanhay endeavorsway [2]. Asway uchsay, it way endslay itselfway
otay uturefay orkway [2]. Orfay instanceway, ethay authorsway ave
hay otnay eenbay ableway otay identifyway away Igpay Atinlay
encyclopediaway, espite day atwhay ightmay eemsay otay ebay anway
importantway eednay orfay oneway. Ethay ethodologymay identifiedway
inway isthay orkway ightmay easilyway ebay appliedway encyclopediasway
inway ethay ublicpay omainday, uchsay asway ikipediaway [7], asway
away enefitbay otay play eakersspay. Ethay eationcray ofway ethay play
ictionaryday ay may alsoway acilitatefay ethay eationcray ofway
additionalway Englishway otay Igpay Atinlay ictionariesday, asway
ellway asway otherway anguagelay otay Igpay Atinlay ic tionariesday
(ethay aterlay ofway ichwhay, ethay authorsway observeway, ashay otnay
eenbay ervedsay byay astpay play iteraturelay). Astlylay, ethay Igpay
Atinlay ictionary day ownay acilitatesfay ethay itingwray ofway ethay
play ationalnay epicway, ethay play anthologyway ofway oetrypay,
andway ethay eatgray play ovelnay. Armedway ithway ethay Igpay Atinlay
ictionaryday, ethay aboveway play ojectspray areway eftlay asway
exercisesway otay ethay eaderray.

Acknowledgementsway

Ethay authorsway ishway otay ankthay otway otnay-osay-anonymousway
eviewersray orfay eirthay eedbackfay andway upportsay. Ethay
Enerablevay Istopherchray Acipay, owhay isway eceivingray ishay bay.
Away. inway inguisticslay isthay ingspray, ovided pray umerousnay
ommentscay onway anway earlyway aftdray. Asway oneway'say eview ingray
atusstay isway oportionallypray ependentday onway ethay umbernay ofway
ass esclay oneway ancay aketay inway away emestersay atway Illiamsway
Ollegecay, Ischray Acipay ountscay asway otway eviewersray. Osethay
owhay isputeday isthay aimclay ancay aketay itway upway ithway imhay.
Ewfay avehay iedtray. Onenay avehay urvivedsay.

Ilewhay itway isway ustomarycay orfay ethay authorsway otay useway
isthay acespay otay extendway away ankthay ouyay otay eykay embersmay
ofway ethay olderway enera tiongay, ethay authorsway avehay osenchay
otay eschewway ethay appreciativeway elodra mamay inway avorfay ofway
away irectday alutesay otay ethay imarypray Igpay Atinlay audienceway.
Otay itway, ethay authorsway ishway otay extendway away ankthay ouyay
otay Englishway-eakingspay ildrenchay orldwideway, ithoutway omwhay
isthay aperpay ouldway otnay avehay eenbay ittenwray.

Aboutway ethay Authorsway

Ianbrayway Irshmanhayway isway away econdsay-earyay aduategray
udentstay inway ethay Oolschay ofway Omputercay Iencescay atway
Arnegiecay Ellonmay University-

way inway Ittsburghpay, Apay. Ehay asway exposedway otay ackbay
angslay anguages lay atway away endertay ageway, asway ehay apidlyray
ecamebay expertway atway ethay "engway-uhgay-uhgay-ishlay" ialectday,
away oprietarypray play-ikelay anguage lay atthay ehay ashay usedway
otay easetay ishay oungeryay otherbray. Oughthay ehay isway otnay
uentflay inway Igpay Atinlay, ehay ashay eenbay eakingspay ethay
anguage lay offway andway onway orfay ellway overway ifteenfay
earsyay. Ianbray oldshay away bay.Away. inway Omputercay Iencescay,
Economicsway, andway Ognitivecay Iencescay omfray Illiamsway
Ollegecay; illway eceiveray ishay astersmay inway Omputationcay, Or
ganizationway, andway Ocietysay omfray Arnegiecay Ellonmay
Universityway; andway isway eekingsay otay applyway otay edicalmay
oolschay isthay allfay.

Onesjayway Aurielayway isway away irstfay-earyay aduategray udentstay
inway ethay Oolschay ofway Omputercay iencescay atway Arnegiecay
Ellonmay Universityway in way Ittsburghpay, Apay. Erhay esearchray
ocusesfay onway ethay ocialsay amificationsray ofway ivacypray, anway
interestway atthay aymay erhapspay avehay owngray omfray away
aumatictray experienceway ithway ethay Igpay Atinlay anguagelay
earlyway inway ifelay. Oughthay eshay ownay acknowledgesway atthay
Igpay Atinlay isway otnay away ecuresay anguagelay, eshay illstay
ondersway etherwhay ildrenchay ecognizeray isthay actfay. Au rielay
oldshay away bay.Away. inway Ociologysay omfray Uway.cay. Erkeleybay
andway away may.Away. inway Omputercay Iencescay omfray Illsmay
Ollegecay.

Osephjayway onoughmcdaywayisway urrentlycay away eachertay ofway
Assicalclay and way Edievalmay Atinlay, asway ellway asway Eekgray
andway Assicalclay Istoryhay, at way Entkay Oolschay inway Entkay,
ctay. Ilewhay originallyway away olarschay ofway aidstay Atinlay
extstay, ehay eservesray away ecialspay aceplay inway ishay urriculum
cay orfay exhaustiveway udystay ofway Atinalay Orcinapay. Ehay isway
especiallyway awndray otay ethay anguagelay ueday otay ishay erfectpay
itchpay abilityway, incesay Igpay Atinlay ovidespray imhay umerousnay
opportunitiesway otay ocklay inway ishay aysway atway anway
enchantingway andway apturousray 110Hz. Oejay oldshay away bay.Away.
inway Usicmay andway Assicsclay omfray Illiamsway Ollegecay, andway ay
may ebay intendingway otay urtherfay ursuepay othbay interestsway
inway ethay oming cay earsyay.

References

[1] W. Idsardi and E. Raimy, "Remarks on language play," 2005.
[Online]. Available: http:// www.ling.udel.edu/ idsardi/ work/
2005lgplay.pdf),

[2] [Online]. Available: Common knowledge [actually available
offline too, so shame on you for looking up this citation]

[3] R. Bavetta, "A pig latin translator," 2008. [Online].
Available: http:// piglatin.bavetta.com/ index.php

[4] Google, "Google igpay atinlay," 2008. [Online]. Available:
http:// www.google.com/ intl/ xx-piglatin/

[5] J. Beech, "Using a dictionary: its influence on children's
reading, spelling, and phonology," Reading Psychology, vol. 25, pp.
19--36, 2004.

[6] wikiHow, "How to speak pig latin," Mar. 2008. [Online].
Available: http:// www.wikihow.com/ Speak-Pig-Latin

[7] Wikimedia, "Pig latin," 2008. [Online]. Available: http://
en.wikipedia.org/ wiki/ Pig [l]{.underline}atin

[8] J. Barlow, "Individual differences in the production of initial
consonant sequences in pig latin," Lingua, vol. 111, pp. 667--696,
2001.

[9] J. Shrager, "Learning lisp," 2008. [Online]. Available:
http:// nostoc.stanford.edu/ jeff/ llisp/ 18.html

[10] H. Chetri and C. Okoye, "The acm "hello world" project," 2008.
[Online]. Available: http://www2.latech.edu/acm/HelloWorld.shtml ˜

[11] P. Erdmann and S.-Y. Cho, "A brief history of english
lexography," 2008. [Online]. Available:
http://angli02.kgw.tu-berlin.de/lexicography/b
[h]{.underline}istory.html

[12] P. Cassidy, "Readme.dic to accompany the gnu version of the set
of files containing the electronic version of the collaborative
international dictionary of english (version 0.46)," Apr. 2002.
[Online]. Available: http:// ftp.gnu.org/ gnu/ gcide/ gcide-0.46/
readme.dic

[13] T. F. S. Foundation, "Gnu general public license, version 2."

[14] H. Alford, "Not a word," The New Yorker, Aug. 2005.

[15] F. Mish, Webster's Ninth New Colligiate Dictionary.
Springfield, MA: Merriam Webster, 1983.

74

SOCIO-ECONOMIC FACTORS AND AUTOMATED STATISTI AL ANALYSIS OF THE
LOLCAT LANGUAGE!!!1!

Dmitry Berenson

Abstract--- OHAI! LOLCAT is a new pidgin language rapdily
{width="2.500999562554681in"
height="2.739332895888014in"}

being adopted for the captioning of animal pictures on the

internet. In this paper, we examine the role of socio-economic

effects in the development of LOLCAT. We also present a new

algorithm called Real-Time Omnibus Feline Linguistics (ROFL)

which monitors key LOLCAT hubs and records LOLCAT

grammar and word-use trends in a readily-accessible database.

Using ROFL has allowed us to track an evolving language that

stands on the brink of suplanting standard languages for cat,

dog, ferret and, most importantly, walrus activity description.

I. INTRODUCTION

LOLCAT is said to have first emerged on the website

www.4chan.org, an online image repository that hosted

weekly cat picture events known as "Caturdays" [5]. Since

it's inception, the language has grown at an exponential

rate, closely correlated with the number of cat pictures

available for public download. The creation of the seminal

LOLCAT hub www.icanhascheezburger.com has unleashed

an explosion (see Figure 1) in the popularity and availability of
captioned animal pictures. A LOLCAT programming language [1] has been
developed, an English to LOLCAT translation website has been created
[3] and a translation of the bible [2] into LOLCAT has been
undertaken. However, while wide-ranging research into the emergence,
popularity, and grammer of LOLCATs [6] [7] has been conducted over
the past several years, an analysis of the root causes and key social
groups that contribute to the development of the language has not been
conducted. Furthermore, a thorough scientific study on the trends
inherent in the burgeoning language has never been completed. This lack
of scientific analysis is largely due to the lack of adequate systems
and algorithms for monitoring and analyzing captioned-picture internet
trends. If we do not take advantage of this unique opportunity to
monitor and study the development of a language, we will be missing a
singular phenomenon in human (and feline) history.

In this paper, we first analyze the socio-economic factors behind the
LOLCAT phenomenon and provide proofs that show its inevitability given
the current state of American society. We then describe the theory and
implementation of the ROFL algorithm and show numerical results
describing recent trends in the LOLCAT language.

II. SOCIO-ECONOMIC ANALYSIS

In this section we endeavor to rigorously analyze the socio-economic
aspects of American society that lead the development and
popularization of the LOLCAT language. Though LOLCAT is now a global
phenomenon, it was originaly developed in the United States, thus we
must

Fig. 1. OMG!!! LOLCAT language explosion!!!

focus our analysis on American society to understand the language's
origins.

Proof: Those who post/view LOLCAT pictures must be fairly affluent.

Lemma 1: LOLCAT pictures are on the internet. Proof: Clearly.

Lemma 2: The internet is a network of computers. Proof: Obviously.

Lemma 3: Computers cost money.

Proof: Everyone knows that.

Lemma 4: Posting LOLCAT pictures takes free time. Proof: Duh.

Thus people posting lolcat pictures have computers and free time and
people with computers and free time are fairly affluent¹. QED

Proof: Information economy creates an increasing demand for cute
animal pictures.

The transition to a globalized information economy has had a
revolutionary impact on American society. The export of manufacturing
jobs overseas and the increasing demand for new technology has created
a need for highly-skilled professionals to create and manage this
technology. In re sponse, American universities and colleges are
graduating an unprecedented number of graduates. While these graduates
generally achieve a higher level of affluence, this benefit comes at a
price. In an increasingly technologized age, affluent people are not
willing to settle for less and demand

¹Note: We do not consider people using computers in public places
such as schools, offices, or libraries, because this would render our
proof invalid.

75

{width="5.553110236220473in"
height="3.001000656167979in"}Fig. 2. LOLCAT macros. (a)
"invisible..." (b) "monorial" © "...ur doing it wrong" (d) "im in
ur..." (e)"...i has them" (f) miscelaneous.

instant gratification. This has lead to a reduction in the number of
children being born to affluent parents because children are generally
considered to require a long and painstaking nurturing period and there
is no gaurantee that one will end up with the child that they want. This
lack of reproduction, however, runs counter to a biological imperitive
to procreate and raise offspring. In response to this lack of offspring,
the psyche of the affluent childless individual is imperiled and seeks
reparation in the less-difficult activity of pet ownership. For some,
pet ownership itself is considered too difficult. LOLCAT pictures can
fulfil the desires of this subset of affluent childless individuals by
allowing access to pictures of others' pets doing particularly cute
things. Thus these individuals can enjoy the positive aspects of
nurturing with none of the downsides. As the economy becomes even more
information driven and technology-centric, this group will increase in
number, thus increasing the demand for LOLCAT pictures. QED

We have thus shown that the demand for cute cat pictures will increase
with the growth of the information economy because of an increase in its
target audience, we will now show why the increase in cat pictures
necessitates the cre ation of the LOLCAT language.

Proof: In order to maintain interest, cat pictures must be captioned
using LOLCAT.

It is a known fact that people quickly tire of content that is too
visceral, i.e. appeals to only the most basic desires. As individuals
effectively overdose on the sacarinity of cute cat pictures, there
must be a cerebral element that involves the prefrontal cortex of the
brain, otherwise the individual becomes bored. Thus some captioning is
necessary to, in effect, "speak" to the reader to keep them
interested. But artirary captions will not suffice because the reader
will become bored by this as well; humorous captions are necessary so
that the reader is consistently "surprised" and thus interested. But
why a new language? The answer lies,

again, in the socio-economic aspects of the target audience described
above. Because most of this audience achieved adolescence some time in
the 90s, they will inherit the dominant humor paradigm of that era,
i.e. sarcasm. Sarcasm is an inherently derogitory humor technique
because it is a way of deriding what is being said through the use of
an exagerated tone of voice, a tone that would presumably be used by
one who actually agrees with the statement being said. Thus latent
sarcasm must be a key component of humorous captions if they are to
appeal to persons who achieved adolescence in the 90s. Indeed LOLCAT
contains a great deal of sarcasm because it is mocking those users of
the internet called newbies (aka newbs or n00bs or even n00bx0rz) who
frequently misspell words and use acronyms such as LOL and OMG. Such
newbies are the victims of constant derision by more experienced
internet users. Thus LOLCAT captures the sarcastic qualities necessary
to sustain the interest of the target audience described above. QED

Thus we have clearly shown how the socio-economic factors of the
modern American economy have contributed to the rise of LOLCAT as an
internet sensation.

III. ROFL ALGORITHM

We now present a method for the analysis of trends in the LOLCAT
language via an automated data-retrieval algorithm termed Real-Time
Omnibus Feline Linguistics (ROFL). The goal of the algorithm is to
track the usage of LOLCAT vocabulary and syntax. The vocabulary we
wish to track is a set of Assinine Acronyms (AAs) that are common in
the LOLCAT lexicon. Examples of AAs are Laughing Out Loud (LOL), Oh My
God (OMG) (note: this AA is usually followed by at least three
exclamation points interspersed with '1's), and Rolling on the Floor
Laughing My Ass Off (ROFLMAO).

The syntax to be tracked is a set of template phrases or "macros"
commonly used by LOLCAT speakers. These are illustrated in Figure 2.

76

The algorithm works via the cutting-edge functionality of the Windows
Application Programming Interface (API). The procedure of the ROFL
algorithm is detailed in Algorithm 1.

Algorithm 1: ROFL Algorithm

Move mouse cursor using WinAPI;

Open Internet Explorer;

Navigate to www.icanhascheezburger.com;

database = [];

while true do

Turn mouse pointer into hourglass;

image = TakeScreenshot();

text = OCR(image);

database = PutInDatabase(database, text);

Turn mouse pointer into arrow;

Position mouse cursor over Refresh button;

Click Mouse cursor;

if Control-C() then

return database;

end

end

Once the algorithm generates a database of LOLCAT vocabulary and syntax
this database can be easily queried to produce statistics about the
prevalence of certain trends in the LOLCAT language. The prevalence of a
certain AA or macro in the database is calculated using Equations 1 and
2, respectively.

{width="3.0001662292213473in"
height="2.4009995625546807in"}Fig. 3. LOLCAT macro prevalence. Matlab
skills, I has them.

{width="3.000999562554681in"
height="2.3726662292213474in"}

_{P(AA) =} e

√_{−πF req(AA)δt}

_q(1)

Fig. 4. LOLCAT AA prevalence. Srsly, I has them.

�

P(M acro) =

_log ^�_{−φF req(M acro)δt}[�]{.underline}_{2 q}(2)

V. CONCLUSION

where δt is the change in time since the beginning of the universe, φ is
the golden ratio, F req(...) is the proportion of the argument in the
database, and q has no meaning whatsoever.

IV. LOLCAT STATISTICS

In this section we discuss recent trends in the LOL CAT language as
determined using ROFL. The data discussed was taken beginning at the
founding of www.icanhascheezburger.com. Examples of each type of macro
considered are shown in Figure 2. Statistics gathered are shown in
Figures 3 and 4 for AAs and macros, respec tively.

From the data displayed in the graphs, it is clear that certain AAs are
rising in popularity while others are going out of style. LOL and WTF
are increasing in popularity while the combersome and blasphemous
ROFLMAO and OMG, respectively, are decreasing rapidly in popularity. In
terms of macros, the "i has them" and "invisible" macros are currently
dominating and miscelaneous is holding strong. "im in ur" has seen a
steady decline since its inception.

In conclusion we have presented a thorough and con vincing analysis of
the socio-economic factors behind the LOLCAT language. We have also
described an algorithm for the automatic collection of LOLCAT data for
later analysis. Our LOLCAT prevalence computation accurately captures
the current trends of LOLCAT AAs and macros and has been used to
generate the informative statistics presented in this paper. KTHNXBYE!

REFERENCES

[1] http://globalnerdy.com/2007/05/28/lolcode-the-lolcat-programming
language/

[2] http://www.lolcatbible.com/

[3] www.lolinator.com

[4] www.icanhascheezburger.com

[5] http://en.wikipedia.org/wiki/Lolcats

[6] Dwight Silverman, "IM IN UR NEWSPAPER WRITIN MAH COLUM :),"
Houston Chronicle, June 2007.

[7] Anil Dash, "Cats Can Has Grammar," May 2007.

77

78

General Case Rendering from Occurring Instances

John Kua

Institutionalized Robotics

Carnage Melon University

Pittsburgh, PA 15213

Email: jkua@cmu.edu

Abstract

{width="6.001832895888014in"
height="4.821832895888014in"}Fig. 1. Number 8 [1]

I. INTRODUCTION

P
^{AUL Jackson Pollock (January 28, 1912 − August 11, 1956) was an influential American painter and a major force in}

the abstract expressionist movement. Pollock was born in Cody, Wyoming
in 1912, the youngest of five sons. His father was a farmer and later
a land surveyor for the government. He grew up in Arizona and Chico,
California, studying at Los Angeles' Manual Arts High School. During
his early life, he experienced Indian culture while on surveying trips
with his father. In 1929, following his brother Charles, he moved to
New York City, where they both studied under Thomas Hart Benton at the
Art Students League of New York. Benton's rural American subject
matter shaped Pollock's work only fleetingly, but his rhythmic use of
paint and his fierce independence were more lasting influences. From
1938 to 1942, he worked for the Federal Art Project.

79

II. THE SPRINGS PERIOD

In October 1945, Pollock married another important American painter,
Lee Krasner, and in November they [REDACTED DUE TO GDFL]

tack the unstretched canvas to the hard wall

$$REDACTED DUE TO GFDL$$

mathematical fractals

$$REDACTED DUE TO GOLF$$

'This is it."'

80