Game Changer - a review

A book review may be a primary source, opinion piece, summary review or scholarly review.
NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: Game Changer - a review

Post by NickFaulks » Wed Sep 09, 2020 7:59 pm

A useful review. I have to say that I did choke on

"Deep Blue, which may or may not have had a little help from its operators".

This ongoing GK propaganda is just nonsense. Yes, Deep Blue's handlers fed in lots of opening lines, which they were entitled to do, but GK still cannot accept the fact that in the game in question he simply played like an idiot.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Paul Cooksey
Posts: 1519
Joined: Fri Oct 21, 2016 4:15 pm

Re: Game Changer - a review

Post by Paul Cooksey » Thu Sep 10, 2020 12:24 am

Idiot seems harsh if we are talking about game 2. But I agree the conspiracy theory now thoroughly debunked.

I enjoyed Game Changer at the start of last year. I suppose a review this long after publication is in itself an indication the book is noteworthy.

NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: Game Changer - a review

Post by NickFaulks » Thu Sep 10, 2020 12:45 am

Paul Cooksey wrote:
Thu Sep 10, 2020 12:24 am
Idiot seems harsh if we are talking about game 2.
I was referring to game 6. I had forgotten about game 2, which makes his play in game 6 seem even stranger.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: Game Changer - a review

Post by Roger de Coverly » Thu Sep 10, 2020 1:01 am

NickFaulks wrote:
Wed Sep 09, 2020 7:59 pm
A useful review. I have to say that I did choke on

"Deep Blue, which may or may not have had a little help from its operators".
Has the Regan software been run against the Deep Blue games, to establish that Deep Blue had engine assistance?

ben.graff
Posts: 98
Joined: Fri Sep 07, 2012 5:53 pm

Re: Game Changer - a review

Post by ben.graff » Thu Sep 10, 2020 8:55 am

It is certainly clear that Kasparov was freaked out and completely lost his cool in his match against Deep Blue. Humans feel emotions while of course machines do not. All agree Kasparov was capable of playing better than he did. Who knows what the result might have been if he had. It could at least have been a closer contest. Clearly were Garry or Magnus to play AlphaZero they could be in the best possible frame of mind and it would not make any difference. We are of course now well past the point where we can blame our failings against the computers on nerves, fatigue or the team around the machine!
Ben Graff
Author of 'Checkmate! Great Champions And Epic Matches From A Timeless Game' 'The Greenbecker Gambit' and 'Find Another Place'

Paul Cooksey
Posts: 1519
Joined: Fri Oct 21, 2016 4:15 pm

Re: Game Changer - a review

Post by Paul Cooksey » Thu Sep 10, 2020 1:05 pm

So a historical digression because I want to make some analogies between Kasparov-DeepBlue and AlphaZero-Stockfish.

I think the standard analysis of the Deep Blue match is that Kasparov went into it expecting to win easily playing anti-computer chess. Programs were at that time so poor at playing slow, closed, strategic positions that they could be outplayed by a GM even from a position that was objective close to lost.

An entirely reasonable assumption given everything seen up to that time and in game 1 of the match. But in game 2 DeepBlue played well in a classic anti-computer position. That no other computer could play that way made a conspiracy theory a human was helping plausible. Kasparov believed it, switched to normal chess, got nowhere and then had a meltdown in game 6.

Other computers did catchup, these days Stockfish 12 sees the positional approach instantly on mid-range hardware. But it really was a watershed moment in that it became clear brute force calculation could get computers to GM level positional play.

I think Game Changer does a good job of the watershed where computers proved they could play with a level of positional understanding beyond humans as well as out calculating us.

I am not entirely convinced that AlphaZero was the strongest computer when it beat a slightly limited Stockfish. Stockfish is improved, I would back it in a rematch. Of could a rematch that Google has no reason to allow. Very much the same situation as with DeepBlue. Kasparov was probably a favourite for a rematch that IBM had no reason to grant.

But in the same way DeepBlue heralded the arrival of the superhuman brute force computers, AlphaZero heralded the arrival of the AI engines. A couple of years on, Leela arguably stronger now on home computer hardware than AlphaZero was back in 2018 with massive IT resources.

I'm not an expert on computer chess, but I became much more interested following Game Changer and by following Matthew Sadler on twitter. It remains an interesting time. I liked Jan Gustafsson's insight that after a decade of everyone at the cutting edge of theory analysing the same positions with the same engines giving the same evaluations, having two types of engine with different preferences has made opening analysis much more interesting. Although I felt for Grischuk when he complained it was a shame neither of them like the Kings Indian!

Post Reply