New Man vs Machine Idea - Chess Forums

RadioRiddles

Jan 8, 2017

0

#1

So a while back i watched this video:

https://www.youtube.com/watch?v=Tnf0W2YxteA

which basically talks about a variation that GM kramnik claims is no longer playable due to the novelty Ne2

But the computer says its about equal.

Then of course recently the stellar game between caruana and nakamura:

And many a GM agreed with caruana that the computer was wrong or at least underestimated the move Nf5.

And these two games got me thinking. What if instead of piece odds, a human was given the winning side of a line which is thought to be positionally refuted according to grandmasters that a computer says is equal. If the human wins then we have bragging rights over the computers that a human's intuition can still defeat the calculating ability of a computer. If the computer wins then it may show some amazing resources which could lead to a resurgence of some lines similar to what happened to the Berlin defence in 2000.

I personally think it's a win/win situation

urk

Jan 8, 2017

0

#2

I've heard Caruana say more than once that the computer is wrong about a line he's played.

I think he actually searches for such positions for his repertoire in order to confound his opponents who trust too much in engine evaluations.

SnatchPato

Jan 8, 2017

0

#3

I think that sounds like a brilliant idea! I am fascinated by the concept that our top GMs still have a deeper understanding of certain positions, purely intuitively, than our super-computer counterparts. To sort of further this experiment, I'm wondering if we could develop a tournament that combines strong GMs and strong engines, and see which "team" is the best in the world.

I remember in the Deep Blue match v Kasparov (the one he lost), Kasparov was complaining that the GMs working with Deep Blue were sort of "guiding" it through the game, but letting it do the brute force calculation still. The point was that the GMs using their insight into positional chess combined with the raw calculation power of Deep Blue was enough to beat Kasparov. Now, whether or not this actually happened is debatable, and not really my point. I just like the idea that human's "guiding" our top engines could result in even more powerful chess playing ability, and it might just bring us closer to, and I strongly hesitate to say this, perfection.

eaguiraud

Jan 8, 2017

0

#4

Good idea, innovative and interesting. Because watching a computer giving odds to a GM is like watching Usain Bolt running against a Ferrari with 3 flat tires.