StephenLawrence - 28 November 2012 11:09 PM
The chess computer, like us, works out that checkmate in 3 would be the consequence of move A and so selects move B instead, it’s working out what would happen if it did A and giving that a lower valuation than doing B, is why it does B. Making move B prevents checkmate in a calculated way. It could have made move A, which means it would have done if it had evaluated that move most highly instead.
(...)
Yes. And you would know that behaviour is the computer evaluating the moves and preventing checkmate. And it’s a machine that’s designed to do that in order to avoid checkmate.
There are many insightful comparisons between the mind and chess programs. This is one. But before somebody says that then chess programs are conscious, one should not forget that the chess computer has no self model. It is not programmed to have a picture of itself in its ‘chess world’.
I would take the example a bit further (based on an example of Dennett): imagine two chess programs playing against each other. They each run in their own part of memory of one single computer, the only ‘contact’ they have is via the interface in which they exchange their moves. Everytime program A wins, a green light is turned on, when program B wins a red light is turned on. Now nearly always the green light is burning.
Then aliens come, who have no idea about chess or our primitive computers. One of the alien researchers, Dr Onestone, wonders why mostly the green light is burning. As he is equipped with the best Quantum-Organic computers of his civilisation, he is able to make a complete representation of the physical structure of the computer. He lets his simulation run, and indeed, nearly always the ‘simulated green light’ burns. So he can be sure he made, in this respect, a correct simulation of the computer. Every other snapshot Dr Onestone makes of his simulation is exactly the same as the factual physical state of the real computer device.
Now the question: does Dr Onestone really understands why mostly the green light is burning?
Dr Onestone hesitates. Would his result justify the astronomical budget he got for understanding this device? In the end, he did not much more than proving that the computer thing device is a physical possible artefact. So he decides to turn to his socio-psychological colleague, Dr Weaver. Dr Weaver then makes interviews with the constructors of the computer device, and after a few days he presents his conclusion to Dr Onestone. This is what he wrote:
The people of this planet sometimes play war games. One of these games is called ‘chess’. It has certain very strict rules, but these rules still allow for practically infinite different ways in which these games develop. The strict rules however made it possible to make computer models of the games, and in this computer two different chess programs play against each other. The bytes in the interface represent the moves the programs make. Everytime program A wins, the green light burns, when B wins the red light burns.
Dr Onestone sighs. So that was it. He realises he will have a heavy day tomorrow, because he knows that his physical simulation in the end brought no insight at all. During a restless night, he suddenly jumps up. One question was not answered by Dr Weaver’s report: why did Program A nearly always win? From his physical model he knows already that program B is as complicated as program A. He looked at his simulation again, but there was no chance to find out what the crucial factor was. Desperately he called Dr Weaver. “Ah”, he says “I did not write it in the conclusion, because for me it was a minor point. Yes, both programs are of about the same level, but program B has one possible chess rule not implemented: it never castles.”
With this final remark, Dr Onestone’s understanding was complete. He realised that without knowing what the physical states mean, one cannot really understand such devices.
