November 2006 Archives

Man, this Natural Language Processing homework is killing me. It's a radical departure from the previous two homeworks. It's not that it's hard, exactly... it's just that this is an assignment inherited from the prof that taught the class last year, and a lot of the documentation links are broken, and even some of the executables appear not to work properly. Many people I've talked to in the class keep getting inexplicable segmentation faults, and it appears to be non-deterministic. The assignment was actually due yesterday, but so many people were having problems with it, the prof granted an extention to the whole class. I plan to submit it tomorrow morning. I think that I'll be able to get it done by then.

We've been doing the group meetings for the students in the Game Design and Development class. It's great to hear all of the students' interesting ideas. Many of the students, naturally, are overly ambitious about what they'll be able to accomplish in six weeks. My constant mantra is "aim low." Aim low. It seems almost unamerican in a way, that phrase, doesn't it? I read that phrase on some game development website - I can't reproduce it here - and it made a lot of sense. It's stuck with me for a long time.

As most of the students in the class are seniors, a lot are out many days for interviews and the like. Now, I didn't interview with anyone when I was a senior - I knew that I was going directly into JET - but I had plenty of friends who did do interviews. But it seems times have changed.

Nowadays, the buzzword seems to be "flyout." As in, "I have a flyout with Electronic Arts next Tuesday." The interview process - for engineers, anyway - is that companies come out to schools to recruit students. Then, if everything goes well, students are "flown out" onsite to see the company in all its cool, hip, sanitized glory. Pure PR. Brainwashing -- I mean, promoting corporate culture.

One thing hasn't changed: signing bonuses. Engineers are an arrogant breed who like to thing they're changing the world; that they're rock stars or sports heroes, and these signing bonuses encourage and perpetuate that notion. One thing has changed, though: the stakes seem to be higher. One student related a story about how his roommate was offered a position at Microsoft, a signing bonus, and also an additional signing bonus if he agreed to not accept a flyout with Google.

Hey kid: feeling lucky?

There's a CNN article about the reorganization of Republican leadership after their defeat in the house and senate. The most noteworthy part of the article, though, is how new minority leader John Boehner characterized the challenge facing the Republicans:

"We need to start by rebuilding the Republican brand," Boehner said. "Republicans need to get back to our core principles and rededicate ourselves to the reform mind-set that put us in the majority 12 years ago."

The Republican brand. I think it's unsettling - and yet extremely appropriate - to characterize a political party as a brand - as a lifestyle choice. It kind of baffles me that we're in a political landscape where it's OK to talk about parties as brands, without any social consequences. I doubt anybody will call him on this, but... how did we get here?

It reminds me of that book No Logo, and the ever-changing face of the Brand.

Here's another quote from Mike Pence, the representative Boehner defeated. According to his webpage:

Mike describes himself as "a Christian, a conservative and a Republican, in that order."

With that context in mind, here's the CNN quote:

"We did not just lose our majority ... we lost our way," Pence said in a Thursday speech to his colleagues. "We are in the wilderness because we walked away from the limited-government principles that minted the Republican Congress."

It always fascinates me how a person's background so dramatically colors the things they say - especially in a formal, official capacity such as this. Phrases like "we lost our way," "walked away" and "wilderness" are extremely evocative of common Christian themes like the dangers of temptation, abandonment, and sin. To me at least.

It would be interesting - and it feeds into my NLP project fairly well - to try to extract whether a given blogger was a member of the religious right or not, based on these collocations. It's definitely feasible.

Weird. I just logged on to check my GMail, and instead of the normal Google GMail logo in the upper-left-hand corner, this appeared, stretched to match the normal logo dimensions:

Has somebody hacked Google? Is it just a bug? Or is it part of some elaborate inside joke? And, most importantly, who are those people? Businesspeople in suits at a conference? Ninjas???

It's still there as I post this. Check it out.

I've been a bit down lately. Have had a lot of things to do - and still have a lot. Too much staying up all night, or nearly so, to get assignments done, or to finish lectures, or whatever. Staying up all night - that always does it. I also had a conversation with my advisor that was a bit of a downer. Or so I thought at the time... still processing it.

He agreed to meet with me after one such all-night session. Monday morning at 9:45, he said, was the only time he could meet me; after that, he was going to be out of town. I needed his advice, so I went.

His office was disorganization incarnate. The Computer Science building is new, though, and I expected that he hadn't yet fully finished his move. Fair enough - there are still lots of unpacked boxes in Heather's and my apartment, after all. I counted about thirty cardboard boxes - some open, some closed, but all stacked haphazardly... leaning, like miniature cardboard Towers of Pisa. Papers strewn about; it was curiously dark. It reminded me of post-earthquake office CCTV recordings regularly aired on Japanese TV. It was almost as if somebody carefully arranged the boxes in such a way to promote the most chaos, like a bonsai of disarray or a Zen of maximal entropy. Really eerie.

"Just move in?" I asked.

"About nine months ago," he replied.

A colleague stuck his head in through the open door. "Hey ther-- whoa!" he balked and surveyed the mess.

"Yeah, I'm going for 'early warehouse'," my advisor joked. Both chuckled, and the colleague went away.

We chatted a bit, and then another colleage passed by. A real motivated type; a go-getter. He greeted my advisor, registered his emotions about the office, and was quickly on his way down the hall before anyone had time to react.

My advisor repeated his comment, loudening his voice for the receding colleague. "Yeah, I'm going for earLY WAREHOUSE!!!!" I heard a muffled chuckle in the distance, over the ringing in my ears.

My advisor advised me well, I think. But our conversation was sobering, in a way that made me, yet again, wonder if I had chosen the right professional path in my life (but then it's seminars like John Holland's and Rahul Sami's that reaffirm my decision - if only temporarily). Compared with many of my peers, my confidence on my life path is positively shaky. They know what they want to do in their life. I'm still figuring it out - and I guess that, on average, I'm seven years older than most of my peers... now that's sobering.

I sometimes imagine that, when I die, a good inscription on my tombstone would be "Here lies John Umbaugh. He never really figured life out." One of my worst fears is that I'll become one of those people my Mom calls a ne'er-do-well. Sometimes I think that the best that I can aspire to is a rare-do-well.

There was a really great AI seminar given today; this time by John Holland, the father of genetic algorithms. That was especially significant to me because I did my undergraduate AI project at Case on genetic algorithms.

There was so much good stuff in this seminar. The topic was about his computer model for the evolution of languages. In a nutshell, he and his team created a model with agents who have - and need to maintain for their survival - a reservoir of different resources harvested from the environment. The examples he provided were shelter, water, and food. Each of the agents could move around, and each had a vision cone: an area of maybe 15 degrees where the agent could sense whatever was within the cone.

He defined an object as salient to an agent if it was present in its vision cone. An interesting feature added to the environment was that, if an object was salient for a given pair of agents, each of those agents was aware of the other's saliency for that object. In other words, agents could detect when they were looking at something another agent was looking at, and vice versa. Holland pointed out that this ability is actually something that is present in the "real" world; wolves and primates, for instance, know when they're eyeing the same thing.

A very interesting note: Holland cited a claim made by some biologists that the whites of the eyes in herd animals (does that term include social animals?) was evolutionary pressured to become white so that other animals (of the same species) could correctly identify what the first was looking at - and, I guess that conferred enough of an advantage that such animals produced more offspring (but one has to wonder about social insects like ants and bees - but then again, they don't have movable eyeballs, and it would cost too much to build moving eyes, so having a contrasting color would be useless - actually worse than useless, after considering the cost).

In any event, Holland also built into his agents a hard-wired mimicry ability. There's lots of evidence that biological organisms possess this ability - a la mirror neurons. Wikipedia defines mirror neurons as:

A mirror neuron is a neuron which fires both when an animal performs an action and when the animal observes the same action performed by another (especially conspecific) animal. Thus, the neuron "mirrors" the behavior of another animal, as though the observer were himself performing the action. These neurons have been observed in primates, including humans, and in some birds.

Interestingly, though, neural networks played no part in the agents' learning. The learning procedure is rule-based.

Finally, Holland's model endows agents with the ability to emit and apprehend utterances - in this case, just a bit string. This is the language of the agents - and the interesting thing here is that the language isn't specified here. The language evolves. In a nutshell, utterances start off as random strings. Utterances are propagated to the next generation if they confer even a slight advantage to the agent over the previous utterances. As time goes on, a language arises. It's an emergent process - just like natural languages.

I've glossed it over a lot here - and there's a fair bit that I still don't understand - but hopefully I can address that next semester. Holland is teaching a course called Complex Adaptive Systems next semester, and I intend to enroll. I hope I can swing it.

There are all sorts of signals all around Michigan's campus today - signals urging people to [get out therer and] vote in today's election. Signs on lampposts; posters on bulletin boards; emboldened, high-point-typeface "VOTE TODAY," all in red, positively imperative capital letters, adorned the front page of the campus newspaper. There were hundreds of blue, circular "Vote!" stickers affixed to the sidewalk at strategic locations.

I imagine people all around the nation are strutting here and there with "I voted!" stickers affixed to their shirts - with the putative implicit message, "Hey there, friend! This is a friendly reminder that today is election day! :)" but with the real implicit message of "What about you, asshole?"

Even in my Knowledge Based Systems class, the prof (whom I admire quite a bit) has gotten into the act. He pointedly reminded all the students (at least those who were US citizens, and actually had the right to vote) that it was everyone's duty, in a functional democracy, to register his or her opinion. A election-time refrain we hear over and over again.

As usual around election time, the media have made sure to cover the low voter turnout phenomenon. Naturally, they generally frame it as the voter turnout "problem." I don't see a problem. There was an NPR segment about young potential voters - particularly the ones most "at risk" to not voice their opinion. "Yeah, I should go out and vote... but I probably won't..." was an often-repeated sentiment. The sentiment of non-voter guilt, - created by our culture, commonly articulated among non-voters.

I didn't vote today. In fact, I've never voted once in my eleven years of votable life (of course, for nearly half of my adult life, I've lived outside of my country -- and who can be bothered with absentee ballots?) - and, while I'm not exactly proud of it, I certainly do not feel guilty or ashamed. Anymore. I feel that I have good reasons for not voting - and I don't have many reasons to begin voting - so I'll leave it at that.

But that's not really the interesting part of all this. The interesting part, I think, is how vigorously we berate non-voters in our culture. We have a right to have a voice, but it's almost anathema if we don't have or choose not to air our voice. That's a little bit weird.

My sister drew this item - http://seto.org/mt-diary/2006/10/open_letter_to.html - to my attention this morning. The author/blogger of Misc. Ramblings, Mr. Dan Seto, has voiced his skepticism about my project. Below is my emailed response - slightly format-edited - to Mr. Seto:

Hello. The other day somebody drew my attention to your entry about my blog project. I wanted to leave a comment, but I see that comments have been disabled for that post. In any case, I still want to give you a response that - hopefully - addressed most or all of your questions. Below is the text of the comment that I composed.

I'll also post this on my blog site in case others might have similar concerns. Thanks!

-John

And here's the text of that would-be comment, also slightly format-edited (I've emboldened my responses):

Hello!

> If he were for real, he probably would have come to my site and got it from the link I have on the right side of the page.

In fact, I don't remember how I obtained your email address, but it's been my experience that often people prefer one over the other - or, in the case of blogs with multiple authors, there are several emails, and I have mistakenly emailed the non-primary contact, if there is one.

> If so, he would know my full name because it across the banner at the top of this page.

Sometimes people use pseudonyms or even complete alternate identities for their blogs. Sometimes it's even a fictional blog with fictional characters. Rather than leave it to chance or to my own judgment, I wanted to have blog authors characterize their own identites to me.

> In fact, if he has come to my site questions three and four are also strange.

Well, if in the case the participant's email was different, I thought that these would facilitate matters. Also, it was my intent that my call for participation would be circulated to other blog authors who might be interested. I know that I've gotten at least a few other interested authors this way.

> In today's environment of epidemic identity theft, question five is none of his business. If he wants to give ranges of ages (e.g., 20 to 29, 30 to 39, etc.), I might answer that. But my actual birth date? Shirley you jest.

Birthdates are in a sense more reliable than ages because, if for example I do testing three months from now, on average one fourth of the ages will be inaccurate by one year (assuming that authors were honest with me in the first place about their ages!). That was my rationale.

> The bottom line is, if this were a sanctioned study of the University of Michigan I would think John would need to get what is known as Informed Consent.

Perhaps indeed I did not provide as much material as I should have on this matter. I have posted my method here. I think this explains in a fair amount of detail why I need the information. As far as safeguarding the information - well, I certainly won't be distributing it to anyone, if that's what you mean. In terms of this being a "sanctioned study" of the University of Michigan - this is simply my class final project, not an "official study."

> And finally, neither his email address nor his website is hosted by the University of Michigan.

I do have a Michigan email (you can search for my last name "umbaugh" on the umich directory for confirmation: directory.umich.edu), but I find it much more convenient to use gmail for my mail, and my personal site for hosting the material.

> In order for me to give this type of information I have to trust and believe John is who he says he is and will use the data for the purpose he states. But since I don't trust unsolicited emails, especially ones that ask these types of question, I will not be answering John. If you are out there John, sorry, but I decline to participate.

OK! Have a good day!

It's snowing today. This time, it looks like it's going to stick around. The professor for whom I do GSI-ing - a native of Sumatra - told me that he found the Ann Arbor winters exhilirating. I've already moved way past exhilirating to acutely annoying. Maybe if I can get some snowboarding or hiking in soon I can curb my seasonal negativism.

That other midterm was today. I was optimistic going into it. Not so optimistic leaving it, however. The good news is that I answered all the questions. Surprisingly, I think that the most difficult questions were the multiple choice ones. It was very difficult to categorize some of the concepts he provided to us. Anyway, I guess I'll get the results next Tuesday. Wish me luck.

Oof, I've been a bit busy lately. Midterms and all that... can you believe that it's time for midterms already?

Last week I took my Language and Information midterm. I felt pretty bad about it... but luckily, as I discovered Tuesday, I didn't do too poorly. It was either a high B or a low A. Still getting an A in the course though, definitely. Tomorrow is my Knowledge Based Systems midterm. This one, I don't think, will be so bad. This one is more critical though, since I didn't do as well on the last project as I had hoped...

I'm reading a really interesting book that I picked up from the Ann Arbor District Library called Flavor of the Month: Why Smart People Fall for Fads. I wanted to include a couple of excerpts that I found really fascinating. First off, an example of a phenomenon that turned out not to be a fad:

Fads and Innovations

Fads can fool us. The sociologist Emory Bogardus made what was probably the first serious study of fads. Each year between 1915 and 1924, he asked about a hundred people to name five current fads. Not surprisingly, he found that most fads did not last long enough to make more than one list; only a few received mention in three successive years. One of these was "men's wrist watches."

Huh? How could Bogardus's respondents have called wristwatches a fad? Looking back, this seems like a foolish mistake. But when they first appeared, wristwatches were considered silly novelties; after all, the pocket watch and chain, suspended from the waistcoat, were as fundamental an element of proper male attire as the necktie. Bogardus's respondents called the wristwatch a fad because they didn't expect it to find lasting favor. And experts agreed. In 1915, the New York Times covered the debate over wristwatches at the National Retail Jewelers' Association convention, and quoted a delegate: "There's some excuse for a woman wearing her watch on her wrist... but a man... has plenty of pockets"; another declared, "The wrist watch will never be a common timepiece." A wristwatch manufacturer explained: "Some of those fellows [are] afraid the sales of wrist watches will hurt the sales of more expensive watches. That's all wrong. A man who has a wrist watch will also have a regular watch."

Here's something revealing about fads and our culture:

People take these campaigns to improve education seriously because they invoke fundamental ideas in our culture, ideas about the possibility and the desirability of change. Our ideas about change provide a cultural foundation for institutional fads. For example, our dissatisfaction with the flaws in today's schools, our conviction that schools could do a better job, and our desire to improve education for future children make us receptive to proposals for educational reform. This is why I can confidently predict that five years from now, those unacceptable, below-grade-level reading scores will be used to justify calls for new solutions to our scholls' problem (although I can't know which proposals will become most popular)....

Although this chapter focuses on American culture and society, I do not mean to suggest that the United States is uniquely susceptible to institutional fads. Rather, we should expect institutional fads to occur in other countries to the degree that their cultures and societies resemble ours. That is, institutional fads will be common in all societies that welcome change, but rare where change is resisted.