In the last twenty years a brand-new window has opened in
every science laboratory, the computer window. Even this ancient writer has bought one and his computer is becoming a friend, though sometimes an exasperating friend. Professional computer-doctors have had to be called in.So far
this particular computer has not been infected by computer viruses, but, as a species they are intriguing. Are they alive? Well, they show many characteristics of life, very similar, for instance to the dengue virus, which, as we
know only too well, is very much alive and kicking.
They can reproduce and propagate (by copying themselves into another computer or floppy disc). They can use the metabolism of their computer host to carry out their own
physical functions , just as does the dengue virus with its chemical functions. They don't have DNA like dengue but they store a representation of themselves in computer code. They respond to their rather restricted environment
(computers, of course). And, to ingratiate themselves with religious people, they were undoubtedly created, if perhaps not Created. Admittedly dengue can live, briefly, outside its normal environment, but there it cannot reproduce.
And it is said that because some computer hackers have a sense of humour, computer viruses are able to evolve.
But that isn't the only way that computers resemble life. They are being deliberately used to produce models of the
way you and I and our friends and enemies actually behave. Let's begin with a very simple one.
Many of us collect in flocks, the million man March, shoals of fish, flocks of birds. Craig Reynolds of Los Angeles wondered how
bird-like agents would behave on screeen. He gave them three simple rules of behaviour, and he called them "boids".
1) Avoid getting too near to other objects, including other boids.
2) Try to keep the same speed as nearby boids.
3) Try to move toward the middle of the boid-mass.
No mention of "Form a flock" you will notice. Rules were very local. But flocks formed every time, even flying
round obstacles like poles in a very life-like manner. In one run a boid hit a pole, fluttered around as if confused and then darted forward to join the flock again.
That led to some interesting thoughts about
"emergence". Would you expect that behaviour to follow those rules, or did it unexpectedly emerge? Let's see what other things can happen when we play games with different rules.
In the part of mathematics known as game
theory there is a well-known scenario called "The Prisoners' Dilemma". Two prisoners in different rooms. Police interrogating about a crime they committed together. No good evidence. Police need confession. The prisoners
know that if neither confess both will go free. But police are cunning. If one confesses he will go free, get a reward, and be immune from prosecution. The other will get convicted and a fine equal to the ratter's reward. If both
rat both get maximum sentence. Interesting?
Prisoner Alleyne is no fool. He knows he can't trust Boyce to keep quiet. The only sensible thing is to tell all. If Boyce is fool enough to keep quiet then Alleyne goes free with a
reward. Sure enough, if Boyce confesses they both go to jail, but at least he, Alleyne, won't be having to pay for it. Logic leads, regrettably, to jail, even though you and I would prefer to see trust and good behaviour,
even in the cells.
In real life it's very unlikely that Alleyne and Boyce met the same day they robbed the bank. Much more probable that they had some previous experience of each other, with dominoes, drinks at the bar,
girl-friends and quarrels. They might be able to trust each other.
Distrust may nevertheless be desirable. Look at how the Cold War and the arms race gave the world forty years of moderate peace. And in the coral reef if I
sidle out of my hole when there's anything biggish around I may end up as lunch.
Now we come to Robert Axelrod of Michigan, a political scientist, for Heaven's sake, about whom I am about to say something good! I think
pre-election polls should be prohibited, for the very good reason that they may influence the election. In spite of the non-scientific inutility of his species Axelrod arranged for a computer test of the Prisoner's Dilemma. Anyone
who wished could take part with a program representing one of the prisoners. And so that Alleyne and Boyce could get some experience of each other, each pair of programs would play 200 games, the programs then being able to respond
to the other's moves. Program Alleyne would be able to assess the reputation of program Boyce.
To start with 14 programs entered. The prize went to a very simple program, ttftt, or Tit for Tat, submitted by Anatol Rapaport of
Toronto. Ttftt's first move was cooperative, and after that it did the same as the other program had done on the move before. Of the other programs six were nice and would never defect first. All of them out-performed the six nasty
ones. Axelrod thought ttftt's win might have been a fluke, so invited players for a second round. Sixty four specially-designed entries failed against ttftt. So nice people can win, especially those nice people who are forgiving,
tough and easily understood.
What a very satisfactory result! My opinion of political scientists has suffered a sea change, though not quite yet to something rich and strange.
Not that such cooperative success is strange to any
of us. Look at the cooperation between insects and the sexual organs of plants, one supplying the tasty grub and the other the chance of having sex with someone other than yourself. Rather more exotically, the things called lichens
live on rock and extract from it useful minerals by a strip-mining fungus,which in turn is grateful for the products of photosynthesis donated by an alga which needs the minerals. Axelrod quotes the cooperative policies of English
and German soldiers in the trench warfare of the first World War, who often spent weeks or more not killing each other until forced into it by a battle organised by some idiot general.
It was even demonstrated later in another
computer model that the policy of Tit for Tat could arise "naturally" in a population of co-evolving individuals. I like to think we have sensible goodness built-in, without the intervention of any god.
