>> But the defeat of the tactics born from this mindset by MCTS at least for
now lay bare the fact that the path to being good at go is actually to
probabilistically sample the phase space of the game and perform impossibly
mindless statistics on the game outcome an enormous number of times ...
This is a rationalist opinion that is attractive because it appears unbiased,
unwilling to make any concessions to human nature and to recognise its
limitations.
And yet- this same opinion overlooks the greatest source of scientific
wonderment in the victories of AlphaGo and family against the best human
players.
Which is to say: that human players play Go (and Chess, and Poker, and Magic:
the Gathering etc) very differently than machines. In particular, human
players do none of the extremely tedious, extremely computationally intensive
maths that computer players have to do. Human players don't perform MCTS,
neither do they train by self-play for many thousands of human-years.
Somehow, humans can play Go and Chess and all manner of board games _without_
having to do any of the hard work that only computers can do reliably. We are
not particularly good at those games- but we can play them well enough that
beating the best of us still takes huge amounts of computational power.
How we do this, why we are even capable of doing this and what other benefits
it confers to us: _that_ is the interesting set of questions. That a big
machine can outperform a human ... we have known this since
ancient times.
There are more things in heaven and earth, Horatio,
Than are dreamt of in your philosophy.
- Hamlet (1.5.167-8)