Originally posted by wormwood 7.Be7 was not played in the game. 7...Be7 was, but then 7...Nxe5 would be time traveling.
That was a mistake on my part. I miscopied. I rechecked the rest, however, and they are all correct. The 7th move on whites part was not part of the variation anyway, so I don't think that it should matter here, I just put it there as a reference point.
Originally posted by FabianFnas Is it a horizon thing? I mean, if this is the last move Fritz have in its combination, then it is good because he black gets a pawn. (Qxc2 would be as good, also giving a pawn.) But if Fritz thought one ply further, (i.e. over the horizon) then he would realize that it was not good for obvious reasons. Somewhere Fritz stops thinking and that's at its hori ...[text shortened]... see easily that the queen at g2 can be taken, but Fritz can't, because he is at the horizon.
Originally posted by amolv06 Ahh, I suppose that makes sense. Thanks.
And I had it set up at 120 seconds.
??? How does that make sense if it's set for two minutes? It's many ply past seeing the loss of it's queen on the next move. Jebus.
Unless you manually set it at a max search depth of one ply regardless of time, which is possible.
He's starting the evaluation at move 7 for black. Move 15 for white, where the queen would be captured, is the 16th ply at that point. It takes a while to get there.
He's not starting the evaluation at the move right before the queen would be captured, when it would be seen immediately.
At some point the engine has to stop and not look at even a single extra move and just evaluate the board. As you get closer and closer to the horizon, the predicted moves are less accurate. The last one is not useful at all, because it doesn't look at even a single reply.
Originally posted by incandenza He's starting the evaluation at move 7 for black. Move 15 for white, where the queen would be captured, is the 16th ply at that point. It takes a while to get there.
He's not starting the evaluation at the move right before the queen would be captured, when it would be seen immediately.
At some point the engine has to stop and not look at even a s ...[text shortened]... accurate. The last one is not useful at all, because it doesn't look at even a single reply.
OH HELL....yes, I missed that second post explaining it.......of course, after getting up to move 15, Fritz would definately not play queen takes pawn, the first post is very misleading, it sounds like that is it's evaluation from that position, not from way back 8 moves at the end of it's list of possibilities and search depth. This entire thread has been a tragic waste of time.
Not really--I think it points out an interesting difference between a variation given by a human and one given by an engine. In a human's variation, every single move is evaluated to the best of that person's ability, and none should be worse than any other, whereas with an engine's evaluation, the moves get worse as you get closer to the maximum search depth (which doesn't matter much, since the engine will reevaluate after each actual move played anyway).
Originally posted by incandenza Yeah, but he listed the variation where it *didn't* get to the next ply. I'm saying if he looks at the next one he should see the difference. It does take about 30 seconds on my machine, so if you're going through very quickly (maybe you set it to analyze the game at 10 seconds per move or something) then it might not show up.
engines run forced variations depth first, so that if there's a Qxg2, it will process the recapture immediately. hence, horizon can't be the issue here. the second, bigger number in the depth field, like 14/42 means the depth of the longest forced variation.
Originally posted by wormwood the engines run forced variations depth first, so that if there's a Qxg2, it will process the recapture immediately. hence, horizon can't be the issue here.
Originally posted by incandenza But what else could be the explanation, then?
I don't know, but my fritz 9 never comes up with the Qxg2 at any point. I suspect he's just doing a copy error over and over again, or something along those lines.
amolv,
1. can you reproduce the error or does it exist only in your initial analysis?
2. did you copy the game manually into fritz or copy&pasted the contents of the pgn into fritz?
3. if you now copy&paste a fresh pgn into fritz (ctrl+v), start infinite analysis (alt+f2), and paste the analysis here after it gets far enough (right click on the engine variations and select 'clip analysis), do you get Qxg2?
Originally posted by amolv06 Heh, I am admittedly not a very good chess player. Just an enthusiast.
And I see what's going on now with Fritz. Thanks for the help.
Sorry, wasn't trying to put you down. You need to recognize that the position after 16. g3 as critical. After Bxe3 what reply does white have? Obviously, fxe3 is disastrous
Even after 16. g3 Rxg3+ 17. fxg3 Qxg3+ 18. Kh1 Qh3+ 19. Kg1(draw??) Bxe3+ 20. Rf2 (forced) Nd4 your position is won. Take your time!!
22 Jul '07 14:10
Is Fritz a little screwy?
Back to Top