Note From France

We are in the town of Calais, France, today, on the North Sea (or the English Channel, depending on your point of view). In World War II the Allied forces attacked here to draw German forces away from Normandy, which is nearby, so we could mount an all-out assault on D-Day. It's cold here, we just went through the town looking for a sweatshirt or something, but no luck.

I will be speaking today at a summer academic program for graduate students and professors who do research in evolutionary computation. My first two talks in Cergy went well -- the second was given at eight forty five in the morning after a conference banquet at a chateau that went on until one in the morning and involved sampling many kinds of wine, so I'd have the say the audience was a little sleepy, but the questions were good and I think it went well. Yesterday we took a train from Cergy back into Paris and from there another hour and a half out to Calais -- it was a long day of traveling.

You may have never heard of evolutionary computing. Sometimes engineers and scientists have problems that are too hard to solve by ordinary mathematical methods. There might be a large number of variables, or there might be a lot of "pretty good but not best" solutions, there might be ways that you seem to be improving your situation but are actually making it worse, well there are lots of things that can make a problem hard.

It turns out that one of the best ways to solve these complex problems is through Darwinian evolution. There are three techniques that are most frequently used in an evolutionary computer program: mutation, crossover, and selection.

You begin by defining a random bunch of guesses at the problem solution. These are just values of variables that can be entered into a formula or tested somehow -- the result of that evaluation is called their "fitness." There are many varieties of evolutionary computation, but all of them include selection, that is, you keep the solutions with the highest fitness, allowing them to reproduce, and throw away the worse ones. This is exactly the "survival of the fittest" that drives natural evolution. Typically you might create a new generation of potential problem solutions by mutating these selected ones, randomly changing some of the values, and you might also use crossover, which is the analog of sexual genetic combination, to combine values from selected "parent" problem solutions. Then you evaluate the new population, select the best, apply mutation and crossover, and go again. Over time the quality of the situations gets better, until you have a solution that meets your requirements. This kind of program is used for everything from searching for new pharmeceutical molecules to analyzing financial data to designing nozzles for diesel engines (I talked to a German guy at the chateau the other night who does that, he and I killed a few bottles of wine while we talked), to any kind of hard problem you can think of.

So today I will speak to a group of EC researchers. I will talk about the power of social interaction for problem solving. I will talk about computer programs that simulate groups of people talking among themselves, trying to figure something out, where everyone is simultaneously a learner and a teacher and over time the population comes to a way of solving the problem. In an evolutionary computation program the population dies out and is replaced by new individuals, in my social optimization programs the individuals persist over time, working together to find solutions to hard problems -- this, too, has been used in many fields, from water runoff engineering to optimizing the media for growing bacteria for research to finding new cures for cancer, there are thousands of applications.

Most of Calais was destroyed in World War II, so the whole town is new, but I'd have to say it was not reconstructed carefully. Everything has a kind of falling-apart feel to it, buildings are poorly made. Last night our shower had no hot water, I mean none, there was a trickle of chilly fluid from the shower-head and that's it. But this morning it came on hot, of course the temperature leaps unpredictably from steaming to freezing but it was possible to shower, at least. Karaoke seems to be a twenty-four-hour-a-day diversion here, and by "Karaoke" I mean "really loud, bad Karaoke" with the doors and windows open. I thought "Feelings" was kind of a bad song when it first came out, but I have now heard the world-record bad version of it, while hoping to fall asleep.

We have now traveled all the way across France by train, from Switzerland to the North Sea. It is a beautiful country. In the Alps we saw smaller family farms but across the country you see acres and acres of agriculture -- there is even a lot of corn here. You pass scattered villages with their red roofs, chateaus and old towers, spacious mansions and tiny farmhouses that are hundreds of years old. Here in Calais the tower downtown was built in the 1300s, can you imagine that? In Annecy I asked my friend, "What is that castle called?" and he shrugged and said, "I don't know, just a castle." The people here have been friendly and helpful, you see a few who seem to feel we should have learned their language before we visited but most of the young people are happy to practice their English a little bit. And the food, wow.

I'm sure Vigilance readers will want to argue among themselves about whether gay people should be treated like there's something wrong with them, and Congressman Weiner's dilemma will go through more chapters during the week. Michelle Bachmann continues to amaze, and New York is moving toward freedom for people to marry without government approval of their sexual orientation. I just wanted to check in from the road. Good Internet connection here, it's not a given everywhere, I will check in when I can.