humans versus computers: making use of time

[ozziejoe]

I have heard it said that each time you double the speed (or thinking time) of a computer, you get about a 50 pnt increase in ELO.


How much ELO boost do you recken a human gets for each doubling of thinking time?

another way to think about this.

if I get 10 minutes for a full game, and my opponent only gets five, how much stronger does my opponent have to be to break even with me. Assume we play 1000 games.


How about if i take 2 hours and my opponent takes 1?



It seems like humans get a rapid increase in ELO at low time controls, but that as time increases, humans don't improve so much. There are limits to how deep we can see, and no amount of time will change that

[Ovyron]

I have heard it said that each time you double the speed (or thinking time) of a computer, you get about a 50 pnt increase in ELO.

Don't forget that ELO starts giving diminishing results after a certain point.

And about humans, it must depend, and I hope other members have a better approximate. As for me, I get a huge ELO boost from just using a fischer clock (at least 3 sec), and a nice base time (at least 5 min), with faster time controls, when I'm in time pressure, I can't stand it and give away won games, so my rating is damaged dramatically.

And my rating goes up until about 20 minutes per game, but I can't use my time properly so my ELO starts going down with more time added after a certain time. That's because I use the same time for my moves whether I have 20 minutes on the clock, or 40, so the extra time only helps my opponent.

I also get bored with too slow games, so, in a game against myself with 40 minutes vs myself with 20, the rating would be about the same for my both selves.

[ozziejoe]

maybe you need to have a word with your 40-minute self: "slow down you 've got plenty of time"

[Ovyron]

maybe you need to have a word with your 40-minute self: "slow down you 've got plenty of time"

Well, I'll make an engine allegory:

Imagine that the engine reaches ply 20 in 3 minutes, and ply 21 in 15 minutes. Then it won't benefit from more time if you already let it examine the games' positions for 3 minutes on the time control; no more time will help until the time control is x5 as big as the max you could give it. If you don't reach this x5, any additional time you give it will be more helpful to the opponent.

Something like that happens to me, it takes me so much time to analyze all the possible replies from my opponent, but then I choose my move and have nothing else to do, so I move. If I didn't see that my move is bad after this, I would require a huge amount of time to spot it, so a little more time would not help, and I end moving too fast on very long games (Otherwise I'd move way too slow, and wouldn't know in what move I should use my extra time.)

Thanks for your interest