[seanhunter]

That was one example of how the author proposes an extremely inadequate analogy. The fact that you can propose an improvement so easily demonstrates this.

It's also very clearly a problem that it ignores player skill, time control and all sorts of other factors that would be basic to any kind of model claiming to have power in predicting chess outcomes.

[jjk166]

> It's also very clearly a problem that it ignores player skill, time control and all sorts of other factors that would be basic to any kind of model claiming to have power in predicting chess outcomes.

This was exactly the point the author was trying to convey. The simplest model of chess with the fewest assumptions is that it is just a random number generator with no dependence on any factors, but this is a bad model, and the "fringe" alien who assumes there is some deeper, hidden structure to the game is correct to do so.

If it helps, consider the alien's sport of glorfball. We've never seen a game played, but we know that approximately 50% of the time the Aberdorfs win, and approximately 50% of the time the Gloophbahorps win. Based on this data, is it reasonable to conclude that glorfball is nothing but a game of chance?

[lmm]

It's reasonable to model glorfball as a game of chance for now, and aim to develop a more detailed understanding either by gathering more data (Can we correlate the occasions when the Aberdorfs win and the occasions when the Gloophbahorps win with anything else? If we can't observe glorfball games directly, we look for ways to find out about them indirectly, or outside conditions that might possibly affect glorfball). It's not reasonable to posit that glorfball results must be driven by cross-referencing the Da Vinci Code against a message written on the back of the Declaration of Independence, which seems to be what the article is advocating for.

[seanhunter]

That's not the simplest model of chess with the fewest assumptions. The simplest model of chess with the fewest assumptions is that white always wins. It's also a terrible model but is significantly better than what the author proposes. It seems very strange that you seem compelled to defend this point.

The author proposes a deliberately terrible model presumably in the hope that he is illuminating a wider point. Sadly I don't think he's doing that.