| I think many comments here miss the point. I’ll try to summarise them and defend the article: * “Chess results are not 50-50” - I don’t think the exact numbers are that important. What’s important is that in the 50-50 model or x-y model the Aliens assume these values to be just a consistent result of an RNG. * “Chess is not random / the players are intelligent / the aliens could investigate who plays against who” - In the parable the Aliens don’t have direct access these variables, just like we might not have the access to some variables in QM experiments and we (and They) could mistake something hard to observe for random (i.e. if the rule is “if there’s an even number of atoms in a 1m radius, do A, if it’s odd, do B” the outcome would be 50-50 and would seem random even if it’s not) * “Theories that are more complicated, predict less or make worse approximations are worth less than simple theories that predict more / better” (in other words Science is about models and approximations, not about truth or “Shut up and calculate”) - I think the author simply disagrees but he also shows why non-mainstream models are worth pursuing and developing - so that we are not stuck in a local maximum with a worse model than the one we could have. Something else concerns me: I don’t know much about QM but if I understand correctly, Bell’s inequality tells us that there are no “hidden variables” and that QM is indeed random. I’d be grateful for any references to articles on this topic that can be understood by non-experts. |
If all you have is data on whether black or white won, then, assuming the 50-50 outcome probabilities, this is a fine model. There is nothing to be gained by anything else.
Where I disagree with the author I guess is to suggest that some other statistically inferior model would be better because, if you had a richer set of observations, it would be shown to be true. To me that scenario is irrelevant at some level, because it's a different scenario (I don't agree with the author's assertion that first move information would somehow decrease predictive information -- it might not improve it, but I doubt it would decrease it).
If you had information on player identities even, you might be able to model ability or something like that, and use that to improve your predictions. But then that relies on the bit of information about which players are which, beyond black and white.
The philosopher Quine argued that models cannot be decontextualized from the observations they are explaining, and I think that is particularly relevant in this case. At some level it doesn't matter whether the Aliens' models are true or not, because they have no information in the scenario that they could use to do anything with.
The 50-50 model (per the author) isn't incorrect, it's just a (assumedly) correct model for the game of chess when all you know is which color won the game. If you were playing chess, and your pairings were random and you couldn't see the board or what pieces were available or where they could go, etc the game would seem very similar to what the aliens were observing. A more "truthful" model is only relevant or useful in the context where there is more data to make it useful.