The Death Of Chess Engines

[Tim Harding]

So where do you find them?

https://play.google.com/store/apps/deta ... omonosovtb

http://tb7.chessok.com/ was the original website, which is what I still use. Back in 2015 you had to pay a small annual subscription but it seems to be free now.

Last year I analysed several 7-man endings on my website using the results from the tablebase:

http://www.chessmail.com/xtras/lomonosov-list.html

I am still using it to examine my 7-man ending against Alan Walton at Reykjavik last month and may add that to the list later.

[NickFaulks]

Thanks, Tim. Very helpful.

[David Robertson]

As I understand it (and I'm inexpert), the tablebases start with mates, and work backwards. Hence, all mating positions arising from three pieces (K v K + x) are fairly straightforward. But thereafter it becomes prodigiously expensive of time/effort beyond a certain point, say seven pieces

[Joey Stewart]

As time consuming as it may seem to a human, fast computers would have no such qualms about analysing millions upon millions of positions, the only thing we are really waiting for is for the processing speed to allow this to be possible.

I am sure there were many who scoffed when computers first started using megahertz processors, and thought there was no way anybody could even need such speed. Now we are in the age of gigabytes, with terrabytes starting to appear more frequently, so extrapolate that another 20 years and I'm fairly certain the numbers will be there and engine development will finally have hit the end of the line.

Obviously, one would hope chess would still exist, as David has pointed out engines are already far above humans anyway and we are still playing but I would imagine the development industry would fall into decline as it would then be possible for consumers to own the best engine and never need a replacement.

[MartinCarpenter]

The number of positions you'd need to formally solve is far past millions. Its a long way past the sort of size of number we can sanely hope to grasp....

The potential sort of solution might be 'opening theory' + 'all 'computer plausible' moves' (say within 0.5 of its best one.). That wouldn't be water tight but it would perhaps be close enough to close the game in some abstract sense.

[Jonathan Bryant]

... I would imagine the development industry would fall into decline as it would then be possible for consumers to own the best engine and never need a replacement.

Maybe, but what percentage of chess 'educational' or ’training' material that’s bought/sold at the moment is based on "need"? I doubt it gets into double digits.

[Stewart Reuben]

The most complex 7 piece endgame is Q+N+K v R+B+N+K. With perfect play by both sides, this is a draw according to the modern Laws. That is because it take more than 75 moves to force the win of material. In total it takes 200 moves by each player to win.
At one British there was K+B+N+N v K+R. We quickly assessed this as a draw. The rook could harrass the bishop. If allowed to sac the rook for bishop, K+2N v K is a draw.

I used to be worried that professional chess would die because computers became so good. Why bother with strong humans? In fact this has not come to pass, which is most heartening.