Moving Blog

I think I will move my blog here because – you have to admit – it’s more reliable than WSO.

Global warming needs a political breakthrough

What has really struck me recently about global warming is how politics is the limiting factor in solving the problem. It seems that the political system is ill-equipped for dealing with something so global and so long-term. As a scientist, I read article after article about how we have the technologies and the economic strategies to solve global warming today, if only the politicians would cooperate (by funding the technologies and implementing the regulatory strategies). As a scientist, I don’t have the tools to even begin to understand what needs to be done to actually get the politicians to cooperate. In other words, I feel helpless when it comes to addressing what seems to be *the* important factor. We need a political science breakthrough rather than a technological breakthrough.

Artificial Intelligence is about Computers making Decisions

All of intelligence – anything observable that intelligent creatures DO – that is, anything by which you can *tell* something is intelligent – is a *decision*. This could be conscious or subconscious, but a decision nonetheless. You could have done something else or said something else. You said “no” to every other option and chose whatever it is you did. I’ve always been interested in this process of decision-making.

Computer science is my methodology – in order to solve problems, I write computer programs. This is how I naturally approach research, this is what I’m good at, this is what I do.

So I think of *intelligence* as *decision making*; and thus I want *computers* to make *decisions*. In fact, AI is all about computers making decisions. I think that is a *very* deep statement. That is what makes AI both powerful and scary. Talking to a computer does not seem so scary; computers making the world’s decisions does seem scary. But a talking computer is one that *makes decisions* about what to say. I can see why using a statistical/probabilistic framework would have been disconcerting for AI pioneers trying to get computers to make the *right* decisions.

Indeed, natural language generation is part of the very *essence* of AI! The very thing that forms the basis of the Turing test! So in some sense it really is “AI-complete”. That’s exciting, but it also means be careful – pragmatically, I need to find research that is tractable. Perhaps a computer that makes decisions for generating natural language about something specific – like Regina’s football database. Or perhaps a computer that makes decisions for generating a specific type of natural language about something general. What might I mean by a “specific type of natural language”? I’ll have to think about that. Or go ask the linguists.

Academics is about intellectual pleasure

I just realized something: academics is pure pleasure. It’s hard to imagine that I’m saying this directly after working on my thesis all night, but it occurred to me while perusing a photography exhibit moments ago. Liberal arts academics is pure, intellectual pleasure. Or call it “quality” from the Zen and the Art of Motorcycle Maintenance perspective. We study ourselves and our world because it’s *interesting*. It feeds the mind.

A guy did a photography project on “body terrains”. He was looking at the similarity of landscapes to human bodily forms (it so happens that my girlfriend was one of these bodily forms, but that’s besides the point). He said there was a lot of “work left to do” in the field of representing landscapes as bodies. I thought, wow. This is just pure intellectual pleasure. At some point in the past, I would have degraded it because it seems so useless from any “practical” perspective. But now I see that it is full of quality. Mind food.

Someone was saying the other day how Scott Lewis, director of the Williams Outing Club, had the best job in the world: he gets to spend his time doing what he loves, wilderness sports — and it’s even healthy! But Joe Cruz has said the same thing about Philosophy; it’s mind food; pure pleasure; intellectual delicacy. I guess it doesn’t contribute to physical health, though.

But it seems clear that this point about pure pleasure gets lost on students starting on day one. What they don’t realize is that all the assignments, the exams, the dissertations, are mostly about finding the people who are most able to intellectualize, and get the most pleasure out of doing so. This is true at least in the framework of applying to graduate school. Sure, your ability to succeed in school is related to your ability to succeed at an intellectual job. But I think I understand a little better now what that grad student at Penn meant about caring more about the research than the location. For him, the pleasure of the thoughts was more important than the pleasures of the friends or the city or whatever else is determined by the location.

I guess it’s just so ironic that students everywhere complain so much about school work, when really it’s all about pleasure.

Today I attended my very last class at Williams College. I’m going to miss this place, for the friends, the fun, and the intellectual pleasure of it all.

Networks vs. Hierarchies as representations of thought and language

My intuition says that creativity is network-based and rational thought is hierarchy-based. Note that a hierarchical tree is simply one form of network or graph.

Also, there is no fundamental difference between graphs and matrices. This could explain why natural language researchers have not been so excited about latent semantic analysis (LSA), which is simply one way of representing a graph in a format that the computer can operate on naturally. This graph represents certain relationships in “meaning” between words.

Finally, an LSA/graph approach may or may not be more appropriate than a hierarchy/grammar approach, depending on the specific problem to be solved. Graphs are less constrained and therefore seem most powerful, though also potentially need a lot more computational resources. Also, human brains appear to be general networks, not constrained to hierarchies, but that may or may not be important to know.

Science and Exponential Growth

This Interesting Thought came when I was thinking about the research I did this past summer. I was working on the problem of automatically finding sentence boundaries in a string of text. The big issue was that in a string of hundreds of sentences, there are on the order of 2^100 ways to place the sentence boundaries — only one of which is correct!! In other words, placing boundaries exhibits exponential complexity.

Our solution to this problem was to take high-probability boundaries as given. Such placements might be incorrect, but usually they aren’t and we were willing to sacrifice this little bit of precision. Because then, we only had to consider multiple possibilities for the boundaries in-between these high-probability boundaries. In mathematical terms, if we split the text into 10 high-probability segments of approximately 10 sentences each, the number of possibilities is only 10 times 2^10: about 10,000 possibilities, easy for a computer to consider. This is FAR, FAR smaller than the original 2^100, which is so big it is essentially impossible to comprehend. 2^100 is approximately the number of words that would be produced if every human being who EVER lived produced an entirely new academic research library every SECOND of their life.

But the point is, it occurred to me that all of science might be framed as a process of isolating a few variables from the vast network of interconnected variables that form the world. It is impossible to analyze more than a few variables at a time because if you assume that most variables affect most others, the number of “effects” (links between variables) exhibits exponential growth (combinations).

From this viewpoint you could say that if you want to make progress in science, you need to find a way to break down the problem you are solving into component parts with few enough interactions. If such small, useful units are found, you can then use them to build back up to the full size of the problem — where building back up has nice linear growth. This makes it possible to actually make progress. You just can’t get anywhere with 2^100 possibilities. Start writing those research libraries…

Inequality

Rick Spalding (a man of wisdom) said that what made the biggest impression on him during the winter study trip to Nicaragua was that the people doing the hardest, most physically demanding (and demeaning) work were getting paid the least.

On the surface, this seems surely wrong. But ultimately whether this is moral seems to come down to the morality of a market economy in general. I think that the basics of such an economy make a lot of sense, but that additional measures are needed to protect the poor. There is some sense of “fairness” which makes it seem like hard physical labor should be rewarded more than skilled mental labor, but if you are doing hard physical labor that no one needs, than why should you be paid at all to do it? This is how the market allocates money “efficiently”.

So the real issue is not the market but that in Nicaragua there are a whole host of other problems, such as illegal maneuvers made possible because of extreme wealth disparities and poverty in general. Not the market but the fact that workers’ rights are not protected.

It’s ironic that the hardest workers are being paid the smallest sums, but what’s not “fair” is that they do not realistically have the option to change their work or improve their working conditions. A free and fair society is one where these opportunities do exist.

Zen and the Art of Motorcycle Maintenance

My mind has been swirling lately with interesting thoughts related to the book Zen and the Art of Motorcycle Maintenance. So as not to be overwhelmed I will try to post them as separate Interesting Thoughts at different Interesting Times.