Making the Brisbane poster:
I returned from Melbourne in 2003 knowing that there would be another meeting of the same organization in 2008 in Berlin, for which I wanted to be ready.  The absence of any cogent objection to what I had presented persuaded me that I was absolutely right.  But everyone I spoke with about it took the attitude, “Ah, but you have no mechanism.  With no mechanism there is no theory.”  I didn’t think a mechanism was that important.  It was obviously true.  I did get some encouragement from someone who said, “Why don’t you make a computer model of it?”  I thought about the billions of base pairs in the human genome and thought it would be impossible to model something that vast.  But eventually I came around and decided to try a vastly simplified model.

It had been years since I had studied Basic computer language, but I thought that was the obvious way to go.  I went down to the nearest college book store to buy one of those little plastic cutouts we used to draw the little Basic symbols and lay out the logic of the calculations.  The fellow behind the counter didn’t even know what I was talking about.  The language of the day was C++ language, which does not offer the “go to” order that is the heart of Basic.  That meant I would have to figure out the logic under C++ rules.  I would also have to teach myself the language.  Learning the language was a little difficult because it was already known that Microsoft would no longer be supporting it.  I suppose there is some other language taking the place of C++.  My understanding is that PHP can be used for a web based interactive program, but I have been unable to get that going to date.

The task of writing the computer code proved to be very difficult.  It was hard on me.  I did not sleep much.  I had every psychosomatic symptom I had ever heard of.  I regret that I didn’t make a list of them at the time.  The nosebleeds were the worst.  Or perhaps the worst thing was the loneliness or the horrors.  But at last the first version of the monster was up and running.  And it indicated that the bigger the population the better.  I was wrong, wrong, wrong.  Everybody else was right.  Diversity rules.  Mix that gene pool, and it will only make things better and better.  I even added a function so that cousins could be excluded from being possible mates either because they were too closely related or not closely related enough.  The results said it was best to be free, free, free.  Anything else reduced fertility.  Months passed.  I had my data.  I knew I was right in principle.  It was just that the mechanism was not there.

It was hard on computers, too.  I burned out four of them.  I suspect that the sheer amount of computation was too much and the processors overheated.  Sometimes I would have a version of the program that was too big for main memory and the program would start to thrash the hard drive.  That slowed it down immensely and resulted in frequent errors, it seemed to me.  Sometimes the computer failed by starting to go at glacial speed at the simplest task.  The last burnout was very embarrassing.  The program started to give the results I had expected.  Ten repetitions gave the same result.  I gleefully wrote up my results and whisked them off to a journal.  The editor was interested and said he was sending it to the referees.  But as I continued to play with the numbers I realized that the results had been spurious.  If you changed one of the parameters by a single unit, something like using 99 instead of 100, the results were the same as they had already been.  The processor was probably damaged just enough to fail with that single set of parameters and fail in a way that I was looking for.  Before I could retract the paper the editor announced that he was not sending it to the referees because is was, “Not of sufficient general interest.”  I think he was telling the truth.  He couldn’t find a referee who was interested. 

Then I finally decided to start over from scratch, only this time I would include chromosomes.  The second program, the final one, was easier to write, mostly because I was far less ambitious.  I left out anything that was not vital.  That meant I could have larger populations modeled, although still not over 20,000.  There was another lucky break.  Online computer gaming had become very popular, and it was computationally very intense.  Modern teenagers were treating their computers with the same enthusiasm that they used to devote to cars, pushing them for performance until they broke.  So at last there were computers with dual processors and special cooling arrangements that could run flat out for hours on end. 

So I tried the new version on a competent gaming computer and voila.  Within a half dozen runs I had the set of parameters I still prefer.  If you look at the results on the Brisbane poster, you may notice that they are not exactly the same as the results on the main page of this site.  I redid all the calculations to be sure it was doing what I said it was doing. 

The Brisbane poster should be a fast read.  Just about all the material has already been presented here, but I go into a little more detail about the program and the way the calculations are done.  I think I was a little more enthusiastic two or three years ago. 

There have been 305 visitors so far.

Home page.