The Power of the Blog

Me. I think. The power of the blog is overestimated at this point. Howard Dean rules in cyberspace, but the majority of people (unsubstantiated opinion) don't care. Cyberfolk are a minorty,a very vocal and active minority, but you can't win an election there. I think Dean is finding that out.

The criticism that I heard is that he really didn't express a position (different than an opinion, but I'll let you figure the difference). His web presence certainly did.

blogging the market

Blogging the algorithm
So, I'm reading this section of this paper and thought of something. First, I think this is very interesting work, in and of itself. Some of the prose is a little overwrought, but there are many good ideas contained within it. I know many people in various management positions would see the blog and this paper as somehow, subversive. I think my boss would hate it, so I made sure to send a link. You never know. They might get it.

Here is my thought. This idea of blogging bubbles seems to be a reasonable analog basis for an optimization algorithm. Similar to a genetic algorithm, but instead of one fitness function there are multiple functions. As the bubble grows from thoughts, to results, to connections, to patterns, the fitness evaluation changes at different levels. Rough outlines, but I think that you could formalize this.

The outcome might be interesting. In much the same way that bubbles are often "false" (i.e., urban legends), the results of the algorithm may be less than optimal. You could calculate urban legends yourself. Is perception reality? Would a social blogging based algorithm result in a different perception? One that isn't true, but has some sort of social value? Do urban legends have social value? Are they cautionary tales for controlling the behavior of youths or the masses?

Rational versus Irrational

Rational versus Irrational
Computers vs. Humans. Up front, I'll say it. I'm an LSU fan. I think they are the most deserving champions. Probably more important is that I'm a computer scientist. I think the BCS functioned exactly as intended. The computer programs are supposed to evaluate each team over the course of the whole season and assign rankings based on that. They did that. When you look at the opponents, schedules and whatnot, you have to admit that OU and LSU had the tougher road to a one loss season. That is the rational way to look at it - judge the whole season, pick the two "best" teams, and let them play it out. The human polls are all about, "What have you done for me lately?" Humans like to consider the most recent events more strongly than those that take place further in the past. That is natural and, frequently, wrong. Although, frequently correct as well. Depends on the situation. (Here is great way to find stuff on a particular subject www.keepmedia.com - here's a bunch on the BCS.)

We like to make predictions. This is what all of science is about. Based on things that have happened in the past, we form hypotheses. We test them, we create models, and if they work out well, we call it a theory (or even a law) and use it to build things for the future. As our understanding of science has grown, we've found that all of our theories are imperfect. Some are quite good, but we always end up with things which are not explainable by that theory or model. Much of this has now come to be lumped as "complexity." Things which can't really be explained except as being inherently inexplicable. We form boundaries and determine what we can about these boundaries, but the predictability of them is beyond us.

This is one of those which isn't predictable. I find it interesting that we look at it as Human's vs. Computers. Humans programmed the computers in a way that did not weight recent games more importantly than games at the beginning of the year. It worked perfectly. It just didn't agree with what the poll voters said. Perfect. It is a different model. If the computer rankings perfectly replicated the human polls, why would we want computer rankings? Neither one really determines a national champion. In sports, they have tournaments. There is no theory, no model, only experimental outcomes. And, in most, there are no repetitions to validate the result. Those are the best. Live in the moment and deal with the event. Baseball is fine and, in many ways, ensures that the "best" team will win (ask the Yankees if that worked in 2003). There is no perfect system for college bowl ranking. If I found one, I wouldn;t tell you about it - I'd go to Vegas and bet on the games. I'd have perfect knowledge of who the best team was in any particular match-up and I'd bet. I'd get rich.

And that is the point of the BCS. Getting rich. Money. That is the bottom line in college football polls, rankings and bowls. Its all about the Benjamins. And we don't really need a poll or a computer to know that.