experiments this year

let’s see what happens

Am I still a racist?

Posted by levi on January 3, 2010

Introduction

On September 28, 2008, I learned that I have a strong automatic preference for whites over blacks by taking an Implicit Association Test at Project Implicit.

The IAT is a test that uses your response time to gauge how closely you associate different stimuli. So, if you’re asked to associate something good with a black person or Hitler, and it takes you a bit longer (a few milliseconds) than it wouldv’e taken you for a white person, the IAT deems that you have an implicit association of goodness for whites over blacks. If IAT people read this, don’t hate me for any simplification or error I make in the description.

UPDATE: After taking the test, I realized that it works a little different than described above. There are about 7 or 8 ‘good’ words like ‘love’ and ‘flower’ and 7 or 8 ‘bad’ ones like ‘hurt’ and ‘agony’. Using the ‘e’ and ‘i’ keys on your keyboard, you have to select pictures of african american faces and european american faces and good words or bad words. If you take it, you’ll understand what I mean.

The Idea

I will take the test again, after having spent 13 months in southern Africa, and a 3 week holiday back in the states. I will then return to Africa, result in hand, for another 11 months to complete my Peace Corps service. I live in a mud hut on a homestead with a Bantu family. If you’re interested to know, they are from the Nyemba, Mbundu, and Kwangali tribes.

Hypothesis

My hypothesis is that, after having spent a year in southern Africa, my score will become more neutral. Since this is essentially a subconscious test, it is very difficult to predict what will happen, but as it stands, my score has nowhere to go but down towards black love.

Results

Well, I took the test, but there were too many errors made to determine a result. I was probably trying to go too fast. I’ll wait a few hours and try again.

Ok, I took it again, twice, just to be sure that the results didn’t change with a retake as some people have claimed. I think those people just get edgy and angry at tests that might imply someone is anything but saintly, but maybe they’re just honestly criticizing the test.

Anyways, I’m happy to report that both times, my result was “Your data suggest a moderate automatic preference for European American compared to African American.” I would have liked to get down to slight, but I’m happy enough with this result. Changing an implicit bias is probably a really hard thing to do.

IAT Screenshot

A screenshot of levi's result on the Black White IAT

I’d like to repeat this experiment again later and see if there are any further changes. Good luck to all of you back here in the states! I’m off on a plane to Africa in 18 hours.

Posted in 2010, By Levi | 2 Comments »

On whether or not the Large Hadron Collider would produce time traveling bosons that nature abhors so much that it would rather give me several million dollars

Posted by kennethmyers on November 7, 2009

Introduction

The Large Hadron Collider has had some awful luck. So much so that even before this week’s absurd baguette incident, internet crazies had already been wondering aloud whether or not God was trying to prevent us from firing the particle smasher up.

But among this ruckus of doom-mongers, the voice of Holger Bech Neilsen incurs my ire less than most. This is, perhaps, because he is “one of the fathers of string theory”.[1]

Neilsen and his colleague Masao Ninomiya have published a few papers[2][3] outlining a far-fetched theory that the successful operation of the Large Hadron Collider might create a particle that would go back in time and interfere with the initial start of the machine. Because such a paradox cannot occur (Neilsen and Ninomiya speculate), the universe is constrained in its options, so that it must select even the most improbable disruptions to the LHC’s mission before it accepts the impossible paradox that would ensue from the machine’s launch.

Neilsen and Ninomiya go on to speculate that the US government’s abandonment of the Superconducting Super Collider was driven by this very same paradox-aversion, and that perhaps even the fall of the Soviet Union was affected by the universe’s need to reduce our political motivation to outshine the Russians by finishing the Super Collider.

Our strange scientific duo follow their diagnosis with a prescription: CERN (the organization that runs the LHC) should create a computer program that simulates a random card draw out of a deck of 10,000,000 cards. One of these cards will be designated the “halt” card. If that card is drawn, CERN should shut down the LHC project and console themselves with the fact that they’ve at least bought some information about temporal paradoxes with all of their billions of dollars, and perhaps avoided the asteroid strike or killer earthquake that otherwise may have been required to shut the collider down.

The CERN people have mostly laughed at this idea (as have I), but I wonder if in some back room they haven’t played the game anyway, just out of curiosity.

Now I’ve thought long and hard about a way that I might be able to stop the LHC myself, because if I could come up with a way in which I was willing and able to stop it, I could conduct an experiment of my own to test the theory of the boson paradox.

Methods

I figured that anyone with a few million dollars could probably launch a decent focused political advertising campaign in Switzerland and get the LHC shut down. Our hypothetical millionaire also would be greatly aided if he could demonstrate that he had set up and won something like Neilsen and Ninomiya’s “card game”, and thus been ordained as a sort of prophet of the universe and charged by God or Fate or what-have-you with the mission of stopping the natural or political calamity that might ultimately be invoked to stop the abhorrent time paradox. The newspapers would eat this up, I think, and the people would listen.

SO – I bought two lottery tickets – one with a jackpot of $39 million and winning odds of 1:175,711,536, and one with a $5.1 million jackpot and odds of 1:25,827,165. I reasoned that with the lesser jackpot and easier odds of the second ticket, I’d approach Neilsen and Ninomiya’s rigorously determined optimal halt card odds of -107, while with the bigger ticket I’d have a chance of winning more money if that’s what would actually be required to shut the LHC down.

Next, I took a solemn vow to spend half my post-tax winnings on a campaign to stop the collider, and I recorded this prophetic video for posterity:

Results & Analysis

I didn’t win.

There are at least a few possible explanations for this:

(1) I would not have been able to compel European voters to shut the LHC down with my money and story.

(2) Something more probable that my winning has been nominated as the LHC stopper.

(3) The universe is stochastic, containing both deterministic and non-deterministic elements, and some of those deterministic (albeit chaotic and therefore pseudo-random) elements are the the machines involved in the selection of my lottery numbers and the winning numbers. There was therefore no statistical wiggle room whatsoever in my lottery play, and my chances were zero all along. (Neilsen and Ninomiya, probably because they had considered something like this, suggested that the card-game emulator be quantum based.)

(4) There are no time travelling paradox-causing bosons.

Conclusion

While I’ve come up with ways to test some of the other possible interpretations of my results that I’ve listed, my favorite hypothesis is #4. I will therefore predict that the LHC will someday soon indeed start.

But if it doesn’t, you can expect me to run another experiment to test interpretation #3.

Posted in 2009 | Leave a Comment »

The ultimate smackdown: exponential technological progress vs. exponential stupidity at the Federal Reserve

Posted by kennethmyers on March 25, 2009

Introduction

In reading Will Durant’s Story of Civilization, one of the things that jumps out at me is the way that the decline and fall of every world empire seems to proceed along the same lines. Transcendental moral values are replaced by ad-hoc relativistic ones, birth rates take a dive, nationalism spikes, and the kings debase the currency by mixing in some worthless metal with the gold.

In thinking about  our own civilization, however, I’m torn. We’ve mostly completed the checklist, and seem to be well on our way to cyclical historical doom, and yet something in me (maybe something arrogant, stupid, and nationalistic) says “But what if this time is different?, What if this truly is “The End of History“?”.

Lewis, in That Hideous Strength, seems to have come to this conclusion by reckoning  that because our wicked empire has become global, there are no barbarians waiting at the gates to give us our much deserved coup de grâce.  (Interestingly, in his novel, Lewis has God step up to the plate and do the eradicating supernaturally in the absence of the much-missed barbarians.)

While this issue is so gargantuan and interdisciplinary that it would be impossible to design an experiment to test it, one facet of the problem that we might be able to explore quantitatively is the debasement of our currency. Inflation (or the deliberate debasement of our currency by printing more and therefore reducing the value of every dollar in circulation) is a kind of hidden tax. It usually runs rampant near the end of a civilization’s life, and it has been running rampant as of late.  As of December of 2008, the Federal Reserve has made the unprecedented move of adopting something called a “Zero Interest Rate Policy” or “ZIRP”, which means that they are now running the dollar-presses at full capacity, in an attempt to produce, as best they can, an infinite flood of money and infinite inflation.

1-purchasing-power-in-20th-century

Enter Ray Kurzweil. Kurzweil is a lunatic or a genius, depending on who you ask, who has been agitating for precisely this move since well before current economic crisis. Why? Because Kurzweil believes that we are living in an unprecedented kind of century, where the explosion of information technology is acting as a DEflationary force which we should be counteracting by printing more money. Kurzweil started making these recommendations in 1987, when he wrote a book predicting the fall of the Soviet Union, the explosive growth of the internet, and the defeat of the world champion chess player by a computer.

6a00d83451ba7969e200e54f81992f8834-800wi

It’s  well-known that technological progress causes deflation. If I can build a better factory that produces T-shirts with half the costs in electricity and labor, you can buy a T-shirt for half the price. In the macrocosm, this means that your money can potentially seem to buy more every year. This effect seems to be more pronounced in information technologies (digital watches used to cost a thousand dollars and now they’re given away free with happy meals), but Kurzweil says that everything is becoming an information technology. In 1987, he predicted that this universal exponential deflation would become so pronounced that people would start to notice it and talk about it . . . in roughly 2009.

So I propose to do the following.

Methods

I intend to calculate inflation like Ron Paul does: by looking at the money supply. Now this becomes problematic, because in order to mask their malfeasance, the Federal Reserve stopped reporting how much money it was creating in 2006 (Yes my friends, it’s getting bad).

I intend to calculate the forces of deflation by looking at the cost of various items (bread, salt, etc.) over the years, and then taking the difference between these prices and the prices one would expect due to the increase in the money supply, and crediting this to the deflationary force of technological progress. Now this is problematic too. There are other factors at play, perhaps most importantly that the U.S. dollar is being used as the world’s reserve currency, and other nations (particularly China) are hording the stuff, and artificially driving up the value of the dollar.

We will hope that these things notwithstanding, I may find some truth. The question that plagues me is this: If we can plot, roughly, the rate at which we are debasing our currency, and the rate at which technology is reinvesting it with value, which curve will win? Does it look like we’re in for a technological deus ex machina, or are we just like every other arrogant dead empire?

Results

The trend of the reinvestment of our currency with value- -be it by Chinese hoarding, technologically driven deflation, or whatever else- -while being so substantial as to keep us from living in a hell world where our dollars would be worth today’s dimes, is so far from being exponential or anything like it that I didn’t even bother to do linear regression:

chart1

Yes, friends, the force of our would-be salvation is the green line. I have to say that I’m glad that I don’t live in a world where the value of the dollar is determined completely by money supply though. Just look at that atrocious blue place at the bottom where we might be.  The following chart depicts just how much of our dollar’s current value can be attributed to our magic deflationary forces:

chart2

There are some caveats that ought to be made.  For one, Kurzweil sees our magic deflationary force really only starting to become a significant issue now, making our historical analysis of trends from 1959 to 2006 quite useless. Another techno-utopian I’ve read, named Virginia Postrel, argues that the Consumer Price Index that I’m using is all out of whack, as it equates a 1905 Ford vehicle with a modern one that goes ten times as fast, or a 1980 computer with a modern one.

My optimistic guess is that they’re both probably right, but it’s not enough to keep me from being terrified at what I see. Personally and practically, what I’ve taken from this study is the idea that winning independence from the monetary system is a better strategy, at least in the short term, than buying into it.

Posted in 2009 | Leave a Comment »

Quantifying complexity in the bible

Posted by kennethmyers on December 16, 2008

Introduction

So Levi’s in Namibia, and he’s got to learn Rukwangali, and the only book around in the language is the bible. Before continuing, let us all pause for a moment to appreciate that Levi’s being forced to read the bible. . . good? Good.

So Levi asked me if I could get my hands on a list of the books of the bible in the order of their simplicity so he could work through it in the most gradual way possible. I looked and looked, and didn’t find anything. Then I decided to make the list myself.  Incidentally, my methodologies have proven useful in my work.

I’m a little disappointed with this experiment because, like the last one, I’m not coming to the experiment through with a body of rival theories that I’ve contemplated, but I’m instead coming to the experiment because it’s an engineering project that’s demanded by the necessites of my social life, and I’m sloppily forming hypotheses and theories after the fact, just so I can call it an experiment. This will change. My next experiment will be better. Promise.

Hypotheses, etc.

SO, I interviewed a Nathan Haydon, Ken Myers the Elder, and Shirley Myers (all church professionals) and got the following bets on what sorts of results I would get:

My hypotheses:  That the easiest book in the bible would be one of the gospels, that Genesis and Jonah would win for the Old Testament, and that almost every Old Testament book would be harder than almost every New Testament book, with the notable exception of Revelation.

Nathan’s hypotheses: Nathan’s thoughts were in large accord with mine, but I coaxed him to make a bet on the simplest book, and he went with the Gospel of Mark.

The Bishop’s hypotheses:  Bishop thinks ”that the Pastoral Epistles will come out easiest, and Leviticus or Deuteronomy will come out hardest (this could be a toss-up with Ezekiel and Daniel)”.

Mom:  Mom thinks like we do, and muses that James might be on the easy side (before I shoot the idea down), and places her bet on Jonah for the easiest.

Methods

Originally, I decided to invent my own really rough metric based on ratio of the number of unique words used in a book to the word count.  Then my boss, Stanton, brought my attention to an existing metric which is basically what I was thinking about, but with the added benefit of a syllable count (to roughly gauge the difficulty of vocabulary) and multipliers that adjust the final score to fit roughly to U.S.  public school system grade levels. This metric is called the Flesch-Kincaid Readability Test, and it looks like this:

 

fk

I got the World English Bible from Project Gutenberg (hey, freeloaders can’t be choosers), and stripped the legalese off of the book headers and footers and fed them one by one through a Flesch-Kincaid program. I recorded the grade level approximates in Excel, and in the end, got this:

Results:

(Yeah, the image sucks. Click here.)

wholebible

Or, to put it in another way:

Psalms 4.67, Job 4.89, Song of Solomon 5.34, Nahum 5.49, Habakkuk 5.68, Proverbs 5.96, Hosea 6.26, John 6.45, Genesis 6.87, 1 John 6.91, Joel 6.93, Jonah 6.98, Zephaniah 7.05, James 7.16, Obahdiah 7.17, 1 Corinthians 7.19, 3 John 7.19, Michah 7.23, Mark 7.4, Luke 7.66, Matthew 7.69, Revelation 7.75, Ecclesiastes 7.77, Isaiah 7.85, Malachi 7.93, Romans 8.25, Galatians 8.51, Haggai 8.74, 2 John 8.86, Lamentations 8.96, 2 Samuel 9.36, Exodus 9.38, Acts 9.42, 2 Corinthians 9.44, Zechariah 9.47, Ruth 9.6, 2 Timothy 9.7, 1, Samuel 9.73, Amos 10.03, Judges 10.47, Philemon 10.74, 2 Kings 10.91, Leviticus 11.03, 1 Thessalonians 11.17 ,Esther 11.5, Philippians 11.51, Hebrews 11.61, 1 Timothy 11.77, 1 Chronicles 11.99, 1 Kings 12.17, Numbers 12.63, 1 Peter 12.78, Jude 12.93, Nehemiah 13.03, 2 Thessalonians 13.3, Jeremiah 13.39, 2 Chronicles 13.57, Daniel 13.71, Ezekiel 13.89, 2 Peter 14.09, Deuteronomy 14.48, Ezra 14.6, Joshua 14.85, Colossians 15.3, Titus 15.6, Ephesians 17.11 

Discussion

So at least some of our general impressions are right. Some of them. I mean, Genesis is easy, like we all thought it would be, and the law books are hard. The Gospels are on the easy side, like everyone knew they would be.  Mom was right about Jonah and James being easy. Nathan was wrong about Mark. Dad was at least right about John being the easiest gospel to read. We were all surprisingly wrong about the New Testament  being a lot easier than the old. We were all also wrong about our picks for the easiest book. But I think the question on everyone’s mind is: PSALMS?!?!?! 

Blame it on the sloppy metric, maybe? Or maybe it’s actually not as hard as we all thought, and we were somehow subconsciously influenced by it’s size. Maybe it actually is formulaic and limited. Doctor Suess-ish, even. “Praise him with X, Praise him with the Y”, “God is good”, “Kill that guy”. I’d love to hear people’s thoughts.

Summary

While there’s  a lot of disturbing surprisingness, the way that the general trend fits my most general assumptions makes me think that the results aren’t all bunk. At the moment, I want to say that they’re probably pretty much right, but that Levi should still probably start with Job, because its narrative form probably has more in common with the kinds of things he’d like to do with Rukwangali literacy than Psalms does.


Posted in 2008 | 3 Comments »

Genetic programming with brainfsck

Posted by kennethmyers on November 12, 2008

 

Introduction

This experiment, like most good experiments, begins with an argument.

flowchart

I was talking to Levi on the phone, a while ago, and told him of my grandiose plans to use the brainfsck (sometimes spelled differently) programming language as the basis for a general purpose genetic programming environment. I figured brainfsck was ideal for this kind of thing, because it only has 8 commands, because hardly anything is invalid code, and because it compiles and runs pretty fast (relatively speaking; I’m used to QBASIC).

Levi seemed to think I was a fool. He opined that it might take longer than my lifetime to get a viable program. I said I wasn’t thinking of doing anything to complicated. He said it didn’t matter. I said “I bet I could make a ‘Hello world’ in a day or so – maybe a week”. He laughed. He said that being able to really use genetic programming in today’s world meant being able to employ really fancy brilliant super high tech complicated algorithms.

So here we are.

Hypotheses and Research Questions

H1: That an English teacher can create a functional general purpose genetic programming system. That it’s not that hard, and that I’ll get a ‘Hello world’ in my lifetime.

H2: That a brainfsck interpreter written in a language I can write in is fast enough to do the job.

Q1: Given that the rate of the progress of my algorithm can be quantified (I talk about how I intend to do this below), what is it?

Methods

I intend to write my breeding and selection program, and maybe even my interpreter, in RapidQ, and roughly perform the following:

(1) Make a lot of programs which are basically just random assortments of the 8 brainfsck commands (+ – < > , . [ ]), while controlling for interminable loops and other hazards as best as I can.

(2) Run them, saving the output to an ASCII text file.

(3) Compare the ASCII text file with a text file I’ve created that says “Hello world”.

(4) Quantify the closeness of program outputs to my ideal text using Levenshtein distances (click the words, the wikipedia page is very easy to understand).

(5) Keep and mutate the programs that do the best, and slaughter the rest and roast them over an open flame on some beach with Mexican beer.

(6) Go back to step 2 and repeat until I’ve got a winner.

Results

DEFEAT.

Levi was right. The algorithms I employed were too simple to get any progress in my evolution.  The Levenshtein Distance, particularly, was to blame. Here’s what happened:

When my program was run, it would first create an arbitrary (determined by me at the onset) number of programs with completely random code. Because I have obviously not solved the halting problem, I put a counter on loops, and any program that looped 100 times was killed on the spot. 

Most programs ran, but didn’t print anything. Out of an initial random set of 10,000 programs, something like 20 of them would print. Sadly, the programs that did print usually printed A LOT. They would do silly things, like looping through the entire character set and printing every third character. This is how Levenshtein was defeated.

Programs that print nothing have a maximum Levenshtein distance of the character count of “Hello World!”: 12. In the many, many times I ran my program, these would always be declared most fit, and all the evolution amounted to was the random shuffling of non-printing code.

Are there ways around this? Absolutely. I could kill all of the programs that didn’t print. I could come up with something of my own that worked better for this project than Levenshtein.

But . . . 

Conclusion

. . . this is exactly what Levi proposed. That the algorithm would have to be studied and refined, and that one couldn’t jump into this kind of thing naively and expect to evolve a functioning program – even one as simple as a “Hello World!”. Still. It is worth noting that genetic programing is SCARY and cool. Watching some program that I didn’t write run on my machine and print crazy characters was exhilarating. I will be back to visit this world again.

untitled-time-0_02_4905

Posted in 2008 | 1 Comment »

 
Follow

Get every new post delivered to your Inbox.