Please look at these videos. It is absolutely fascinating. GMs have no other option than to learn from AlphaZero's style of play.
youtu.be/akgalUq5vewLygo
http://mlwi.magix.net/bg/chessvar.htmSo amazing... how many years till it starts to build terminators?
Its really amazing how A0 could beat a crippled opponent with 1000X bigger computer and turning off the opponents opening book.
When it comes to chess computer software, the hardware needs to be equal. Period.
Fake News.
No opening book, no endgame table bases, fixed 1 min/move, basic hardware... not a fair match indeed.
Nevertheless interesting game concepts, especially the two wins with black.
If you compare the number of operations AlphaZero calculates per second to Stockfish (even in this so-called weakened state) the number are massively in Stockfish's favour. And by the way Alpha Zero did not have an opening book (and it had only 4 hours of self-learning experience). Please stop holding these prejudices, everything will be proven in the "equal" match that will eventually be held.
What does it matter regarding...
the restrictions/limitations placed on SF were similar to a 100 meter dash where one runner had to wear a bag around their legs and carry and 200 lb. sack of potatoes.
SF should have won anyway! For crying out loud AlphaZero did not even posses an evaluation function. We have no idea how it choose it's moves.
Arbiter !
#5 Does it mean that stockfish calculate much more than a 2500 rated engine?I am not sure on that
This is "fake news" that A0 did not have an "opening book".(It did not have one because the self teaching logarithm does not require one.) However, A0 had full access to all the 100's of millions self taught games it played. It did not create an "opening book" in the traditional sense of current programming by humans. After it learned how to play it was able to access it's previous games but how much so has not been released. It has no evaluation function. It ended up selecting a very few set of openings that it deemed has giving the best results. It's selection of moves is based on logarithms not requiring traditional evaluations or past game searches. . It played a few selective openings and took full advantage of the many mistakes, inaccuracies and sometimes outright blunders made by SF. Without it's opening book it's surprising SF made any draws at all. Many of the serious mistakes were made in critical positions where the 1min/move restriction put SF at a severe disadvantage.
To post a video from "
Chess.com" is low
Stockfish did not have an opening book?? How come it played correct theory then? In that case A0 had an undue advantage, because it knew which opening lines were good, whereas Stockfish had no knowledge at all. This can't be right.
