Nezhmetdinov - Chernikov - 1962 - Queen Sac

[Lyudmil Tsvetkov]

and one more:

[pgn][Event "Blitz 10m+2s"]
[Site "Microsoft"]
[Date "2016.12.14"]
[Round "?"]
[White "Komodo 10.1 "]
[Black "Komodo 10.1, owner"]
[Result "*"]
[Annotator "owner"]
[SetUp "1"]
[FEN "5k2/1p1r1p1p/r2N1P2/p1P1P1p1/3B4/1P5P/4q1P1/1R1R2K1 w - - 0 1"]
[PlyCount "20"]
[TimeControl "600+2"]

{512MB, OWNER-PC} 1. Re1 {-0.82/20 4} Qd3 {-0.96/20 4} 2. Red1 {-0.90/20 2} Qc2
{-0.91/21 3} 3. Rdc1 {-0.90/20 3} Qg6 {-0.83/20 3} 4. Rf1 {-0.63/18 2} Qc2 {-0.
62/20 3} 5. Rfc1 {-0.29/21 4} Qg6 {-0.75/21 3} 6. Rf1 {-0.60/21 3} Ra8 {-0.70/
21 5} 7. c6 {-0.54/21 4} bxc6 {-0.49/22 1} 8. Bc5 {-0.55/23 2} Kg8 {-0.34/21 3}
9. e6 {-0.26/22 2} Rxd6 {-0.36/21 3} 10. Bxd6 {-0.25/22 2} Qe4 {1.20/18 1} *

[/pgn]

see how dangerous the white position actually is! suddenly the score changes from negative to +200cps.

still have note seen a single black win the way I check, but an awful lot of white wins.

[yanquis1972]

you seem to be playing at very fast TC or using limited depth searches it simply won't yield productive results when precision is required. for example 9...Rxd6?? loses, but fe should hold.

[d]r5k1/3r1p1p/2pNPPq1/p1B3p1/8/1P5P/6P1/1R3RK1 b - - 0 9

[Milos]

you seem to be playing at very fast TC or using limited depth searches it simply won't yield productive results when precision is required. for example 9...Rxd6?? loses, but fe should hold.

[d]r5k1/3r1p1p/2pNPPq1/p1B3p1/8/1P5P/6P1/1R3RK1 b - - 0 9

Gee, don't you see, it's like ARB, he is just trolling. He starts selfplay, and then when he sees fail-low during black's thinking time he forces the engine to play, therefore you see ridiculously low depths with huge score change. He is totally fake, like "Father" or ARB. I really don't get it why ppl even bother...

[yanquis1972]

perhaps; but i think instead he sees the winning potential for the white pieces & naturally but mistakenly evaluates it as winning as a result. in human games theres a lot of potential for white, but against engines (at least with enough time), its just a draw.

[Milos]

perhaps; but i think instead he sees the winning potential for the white pieces & naturally but mistakenly evaluates it as winning as a result. in human games theres a lot of potential for white, but against engines (at least with enough time), its just a draw.

There is zero white potential, white can only lose. The only way for white to preserve draw is perpetual or repetition.

[Lyudmil Tsvetkov]

hi lyudmil, been a bit awol w/ a bad cold but good to see theres still interest/debate here.

i have to say tho i agree w/ bram, i simply don't think you are letting komodo reach sufficient depth. i certainly havent done deep analysis, but letting it reach even what i'd consider fairly low depth from these positions (pretty fast TC on quad core PC, say), i don't think you'll see these + evals for white. i can at least say that i never did, even after playing (according to the machine) objectively inferior moves to reduce material & hopefully prove a draw more speedily. granted i was almost never able to suggest superior alternative moves or come up with a plan myself which is where your & others input is extremely helpful.

but, doing my best to execute (with engine help) any plans, even allowing them with black, the machine would usually find an escape route. its definitely a troublesome position but given enough depth i still believe its an easy draw for an engine w/ black.

will try to do some engine v engine from these positions myself to show you what i mean; or perhaps ill be surprised myself.

hi John, please try the way I suggested: check by renewing the search on every ply, I do that in Fritz gui for example by pushing the go button, but I guess every gui has its own force command; so simply push the go button for white as well as black, for say 3 or 4 seconds, every 3 or 4 seconds, and then you will be surprised.

[Milos]

hi John, please try the way I suggested: check by renewing the search on every ply, I do that in Fritz gui for example by pushing the go button, but I guess every gui has its own force command; so simply push the go button for white as well as black, for say 3 or 4 seconds, every 3 or 4 seconds, and then you will be surprised.

Lol, imagine there are ppl like you who even play CC like that and never realize why they can't even reach 2000Elo
You must be thinking you invented some super search method. Have you ever wondered why no-one tried to implement such a genius method of yours in actual search?

[Lyudmil Tsvetkov]

you seem to be playing at very fast TC or using limited depth searches it simply won't yield productive results when precision is required. for example 9...Rxd6?? loses, but fe should hold.

[d]r5k1/3r1p1p/2pNPPq1/p1B3p1/8/1P5P/6P1/1R3RK1 b - - 0 9

this was just an illustration how dangerous the white position is; in any case, white won, not black.

any shorter search that is renewed on every ply is infinitely more reliable than even a very long search from the root position; for example, SF will see mate in 3 in 0.0001 seconds, while the same mate in, say 27, will never be seen by the engine. it simply does too many formulistic things pruning away good moves, like following pv, when it should not, pruning away good moves due to other formulas or bad move ordering, etc.

so, of course, a shallow search that is renewed at every ply is much more accurate.

please do the way I suggest: push the go button, let the engine think 3-4 sec., then push the go button again, let the engine think for 3-4 sec. again, and push go again, etc.

in this way, you can even check different lines.

[Lyudmil Tsvetkov]

btw., no alternative to Rd6, otherwise white pushes e7.

[Milos]

any shorter search that is renewed on every ply is infinitely more reliable than even a very long search from the root position; for example, SF will see mate in 3 in 0.0001 seconds, while the same mate in, say 27, will never be seen by the engine. it simply does too many formulistic things pruning away good moves, like following pv, when it should not, pruning away good moves due to other formulas or bad move ordering, etc.

so, of course, a shallow search that is renewed at every ply is much more accurate.

please do the way I suggest: push the go button, let the engine think 3-4 sec., then push the go button again, let the engine think for 3-4 sec. again, and push go again, etc.

in this way, you can even check different lines.

Oh you poor not-understood genius. Maybe my 1000 euros offer will help you realize what a genius you are. Just accept it and play, don't be shy, you can't lose, you have your LRB method