yeah, now SF is clear top.
It's a shame that nTCEC will now be using the old SF and other older engines. too late to change anything now!
But wait! ntcec is about entertainment isn't it? it's not about which one is best.
so that's excellent!!!!!!
(just kidin)
18 days from SF4 release and about ~30+ ELO gain!
Moderators: hgm, Rebel, chrisw
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
-
- Posts: 10416
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: 18 days from SF4 release and about ~30+ ELO gain!
It seems to me that most of the progress that stockfish make is thanks to search improvements.
I feel that stockfish's search is still weak part of it otherwise stockfish's developement version could solve the following position fast
[D]6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
I can accept being twice slower than most top programs but stockfish developement version seems to me more than 10 times slower than most top programs after the stockfish team decided to reduce dangerous moves because unfortunately it gave some elo improvement.
Note that my last patch that passed both stage I and stage II correct the problem but marco is going to test a simpler patch that helps to find Rb3 faster than the developement version but still in the order of 10 times slower relative to my patch(and it is not that my patch could help stockfish to find Rb3 faster than the best top programs so I do not feel that I extend too much).
Note that my intuition tells me that if we want stockfish to be strong against weak opponents then we need to have less positions when stockfish is significantly weaker than them in tactics.
Maybe my intuition is wrong but my intuition tells me that the patch of not reducing dangerous moves has to be counter productive
if we test against opponents that are 200 elo weaker that is important for rating lists.
I feel that stockfish's search is still weak part of it otherwise stockfish's developement version could solve the following position fast
[D]6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
I can accept being twice slower than most top programs but stockfish developement version seems to me more than 10 times slower than most top programs after the stockfish team decided to reduce dangerous moves because unfortunately it gave some elo improvement.
Note that my last patch that passed both stage I and stage II correct the problem but marco is going to test a simpler patch that helps to find Rb3 faster than the developement version but still in the order of 10 times slower relative to my patch(and it is not that my patch could help stockfish to find Rb3 faster than the best top programs so I do not feel that I extend too much).
Note that my intuition tells me that if we want stockfish to be strong against weak opponents then we need to have less positions when stockfish is significantly weaker than them in tactics.
Maybe my intuition is wrong but my intuition tells me that the patch of not reducing dangerous moves has to be counter productive
if we test against opponents that are 200 elo weaker that is important for rating lists.
-
- Posts: 2283
- Joined: Sat Jun 02, 2012 2:13 am
Re: 18 days from SF4 release and about ~30+ ELO gain!
Very interesting, Uri. When was this (counter productive) patch introduced?Uri Blass wrote:It seems to me that most of the progress that stockfish make is thanks to search improvements.
I feel that stockfish's search is still weak part of it otherwise stockfish's developement version could solve the following position fast
[D]6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
I can accept being twice slower than most top programs but stockfish developement version seems to me more than 10 times slower than most top programs after the stockfish team decided to reduce dangerous moves because unfortunately it gave some elo improvement.
Note that my last patch that passed both stage I and stage II correct the problem but marco is going to test a simpler patch that helps to find Rb3 faster than the developement version but still in the order of 10 times slower relative to my patch.
Note that my intuition tells me that if we want stockfish to be strong against weak opponents then we need to have less positions when stockfish is significantly weaker than them in tactics.
Maybe my intuition is wrong but my intuition tells me that the patch of not reducing dangerous moves has to be counter productive
if we test against opponents that are 200 elo weaker that is important for rating lists.
By all means, more testing should be done against such weaker opponents to uncover weaknesses. The computer chess world is not limited to SF, Komodo, Houdini, Rybka etc.
Such weaknesses can also detract from SF's value as an analysis engine... OR maybe this could be good news for the commercials. If the engine offers crappy analysis, what benefit will it offer me if it can crush Houdini ?!
Regards,
CL
-
- Posts: 10416
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: 18 days from SF4 release and about ~30+ ELO gain!
some days after the release of stockfish4carldaman wrote:Very interesting, Uri. When was this (counter productive) patch introduced?Uri Blass wrote:It seems to me that most of the progress that stockfish make is thanks to search improvements.
I feel that stockfish's search is still weak part of it otherwise stockfish's developement version could solve the following position fast
[D]6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
I can accept being twice slower than most top programs but stockfish developement version seems to me more than 10 times slower than most top programs after the stockfish team decided to reduce dangerous moves because unfortunately it gave some elo improvement.
Note that my last patch that passed both stage I and stage II correct the problem but marco is going to test a simpler patch that helps to find Rb3 faster than the developement version but still in the order of 10 times slower relative to my patch.
Note that my intuition tells me that if we want stockfish to be strong against weak opponents then we need to have less positions when stockfish is significantly weaker than them in tactics.
Maybe my intuition is wrong but my intuition tells me that the patch of not reducing dangerous moves has to be counter productive
if we test against opponents that are 200 elo weaker that is important for rating lists.
By all means, more testing should be done against such weaker opponents to uncover weaknesses. The computer chess world is not limited to SF, Komodo, Houdini, Rybka etc.
Such weaknesses can also detract from SF's value as an analysis engine... OR maybe this could be good news for the commercials. If the engine offers crappy analysis, what benefit will it offer me if it can crush Houdini ?!
Regards,
CL
Here is the relevant patch that made stockfish significantly slower in solving the diagram position but better against previous versions.
Author: Gary Linscott
Date: Thu Aug 29 23:02:18 2013 +0200
Timestamp: 1377810138
Enable LMR for dangerous moves
Passed both short TC
LLR: 2.96 (-2.94,2.94)
Total: 5598 W: 1250 L: 1125 D: 3223
And long TC
LLR: 2.97 (-2.94,2.94)
Total: 16441 W: 3102 L: 2912 D: 10427
bench: 4620975
-
- Posts: 6442
- Joined: Tue Jan 09, 2007 12:31 am
- Location: PA USA
- Full name: Louis Zulli
Re: 18 days from SF4 release and about ~30+ ELO gain!
The latest Stockfish is indeed much slower here than Critter:Uri Blass wrote:It seems to me that most of the progress that stockfish make is thanks to search improvements.
I feel that stockfish's search is still weak part of it otherwise stockfish's developement version could solve the following position fast
[D]6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
I can accept being twice slower than most top programs but stockfish developement version seems to me more than 10 times slower than most top programs after the stockfish team decided to reduce dangerous moves because unfortunately it gave some elo improvement.
Code: Select all
Stockfish sig-3864419 64 SSE4.2 by Tord Romstad, Marco Costalba and Joona Kiiski
Searching: 6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
infinite: 1 ponder: 0 time: 0 increment: 0 moves to go: 0
1 -0.44 00:00 1738 Qe1+ Rf1
2 -0.97 00:00 3008 gxf6 Rxf6 Qb1+ Rf1
3 -1.05 00:00 4219 gxf6 Rxf6 Qb1+ Rf1 Qc2 Qxc5 Rxb2 Qxd4 Rxa2
4 -1.05 00:00 15433 gxf6 Rxf6 Qb1+ Rf1 Qc2 Qxc5 Rxb2 Qxd4 Rxa2
5 -0.87 00:00 21889 Rf7 Qxc5 gxf6 Rgg2 Qe1+ Rf1 Qe3+ Rgf2 Qg3+ Rg2
6 -1.74 00:00 36015 d3 Qc3 Rf7 Rgg2 gxf6 Rxf6 Rxf6 Qxf6
7 -1.47 00:00 41610 d3 Qc3 Rf7 Rgg2 gxf6 Rxf6 Qe3+ Rff2 Rxf2 Rxf2
8 -1.45 00:00 68464 d3 Qc3 Rf7 Rg3 a3 b3 Rxf6 Rxf6 gxf6 Qxf6
9 -1.62 00:00 140252 Qb1+ Rf1 Qd3 Re1 Qxa3 bxa3 Rb1 Rxb1 Bxb1 Rxg7+ Kf8
Kf2 Bxa2
10 -1.62 00:00 192966 d3 Qc3 Rf7 Rg3 a3 b3 Rxf6 Rxf6 gxf6 Qxf6 Qe1+ Kg2
Qe4+ Qf3 Kh7 Rh3+ Kg8
11 -1.47 00:00 282211 d3 Qc3 Rf7 Rg3 Rf8 Rf1 Qe8 b3 axb3 axb3 Rxf6 Rxf6
gxf6 Qxd3 Qe1+ Kg2
12 -1.60 00:00 722527 d3 Qc3 Rd7 fxg7 Rxg7 Rg3 Qd4 Qxd4 cxd4 Kf1 Bf7 Rxg7+
Kxg7 b3 axb3 axb3 c5 Ke1 Be6
13 -1.84 00:01 1281K d3 Qc3 a3 b3 Rf7 Rg3 Rxf6 Rxf6 gxf6 Qxf6 Qe1+ Kh2
Qe2+ Rg2 Qh5+ Kg1 Qd1+ Qf1 Qxf1+ Kxf1 Kf7 Ke1 Kf6
Kd2 Be4
14 -1.84 00:01 1357K d3 Qc3 a3 b3 Rf7 Rg3 Rxf6 Rxf6 gxf6 Qxf6 Qe1+ Kh2
Qe2+ Rg2 Qh5+ Kg1 Qd1+ Qf1 Qxf1+ Kxf1 Kf7 Ke1 Kf6
Kd2 Be4
15 -1.96 00:01 2947K d3 Qc3 Rd7 fxg7 Rf7 Re5 Qd4 Qxd4 cxd4 Re6 Bh5 Rd2
Rxg7+ Kf2 Rg6 Rxg6+ Bxg6 Ke1 Kf7 Rh2 Ke6 Kd2
16 -2.08 00:02 4845K d3 Qc3 Rd7 fxg7 Rxg7 Rg3 Qd4 Qxd4 cxd4 Kf1 Bf7 Rxg7+
Kxg7 b3 Kg6 Ke1 axb3 axb3 Be6 Kd2 c5 Rf3
17 -2.06 00:02 5360K d3 Qc3 Rd7 fxg7 Rxg7 Rg3 Qd4 Qxd4 cxd4 Kf1 Bf7 Rxg7+
Kxg7 b3 Kg6 Ke1 axb3 axb3 Be6 Kd2 Bf5 Rf4 c5
18 -2.18 00:02 10336K d3 Qc3 Rf7 Re5 Qg4+ Rg2 Qd7 fxg7 Rf6 Rxc5 Rd6 Qd2
Qxg7 Qf4 Rf6 Qb8+ Rf8 Qe5 Rf6 Rxc6 Rf1+ Kxf1 Qxe5
Rcxg6+ Kf7
19 -2.20 00:03 12104K d3 Qc3 Rf7 Re5 Qd4 Qxd4 cxd4 Rg5 Bh7 Rfg2 g6 Kf1
Rxf6+ Ke1 Kf7 Rh2 Bg8 Kd2 Rf1 Ra5 Kf6 Rc5 Be6 Rxc6
Ke5 Kxd3 Rf3+ Kc2
20 -2.10 00:03 15502K d3 Qc3 Rf7 Re5 Qg4+ Rg2 Qd7 fxg7 Rf6 Rxc5 Rd6 Qd2 a3
Ra5 axb2 Ra8+ Be8 Rb8 Re6 Rxb2 Qd4+ Qf2 Re1+ Kh2
Qd6+ Rg3 Qh6+ Rh3 Rh1+ Kxh1 Qxh3+ Kg1 Qg4+ Kf1 Kxg7
Qg2 Qxg2+ Rxg2+ Kf6 a4 Bf7 Rf2+ Ke7
21 -1.88 00:06 44220K d3 Qc3 Rf7 Rg3 Rf8 Rff3 Rf7 Rxd3 Rxf6 Rd8+ Rf8 Rxf8+
Kxf8 Qf3+ Qxf3 Rxf3+ Ke7 Rg3 Kf6 Kf2 Bf7 Ra3 Bxc4
Rxa4 Be6 Ra6 Bd5 Ra5 c4 Ra7 g5
22 -1.92 00:07 48941K d3 Qc3 Rf7 Rg3 Rf8 Rff3 Rf7 Rxd3 Rxf6 Rd8+ Rf8 Rxf8+
Kxf8 Qf3+ Qxf3 Rxf3+ Ke7 Rg3 Kf6 Kf2 Bf7 Ra3 Bxc4
Rxa4 Bd5 Ra5 Ke5 Rxc5 Kd6 Ra5 g6 a4
23 -1.25 00:10 74175K d3 Qc3 Rf7 Rg3 Rf8 b3 a3 Rf1 Rf7 Rff3 Kh7 Rh3+ Kg8
Rxd3 Rxf6 Rd8+ Kf7 Qd2 Qg4+ Qg2 Qxg2+ Kxg2 Be4+ Kg1
Rg6+ Kf2 Rg2+ Ke3 Bb1 Rd7+ Ke6 Rc7 Kd6 Ra7 Rxa2 Rxg7
24 -1.88 00:12 91075K d3 Qc3 Rf7 Rg3 Rf8 Rf1 Rf7 Rff3 Qe2 Re3 Qd1+ Re1 Qh5
Re8+ Kh7 Ree3 Qd1+ Re1 Qh5 fxg7 Rxg7 Rg2 Qf3 Rf1
Qe3+ Rff2 Qd4 Qxd4 cxd4 Rh2+ Bh5+ Kf1 Kh6 Rf6+ Rg6
Rxg6+ Kxg6 Ke1 Kg5 Kd2
25 -1.64 00:14 103718K d3 Qc3 Rf7 Rg3 Rf8 Rf1 Rf7 Rff3 Qe2 Re3 Qd1+ Re1 Qh5
Re8+ Kh7 Ree3 Qd1+ Re1 Qh5 fxg7 Rxg7 Rg2 Qf3 Rf1
Qe3+ Rff2 Kg8 Rg4 Qe6 Rg5 Qe3 Rg2 Qd4 Qxd4 cxd4 Kf1
Bh5 Rxg7+ Kxg7 Ke1
26 -1.41 00:16 115648K d3 Qc3 Rf7 Rg3 Rf8 Rf1 Rf7 Rff3 Qe2 Re3 Qd1+ Re1 Qh5
Re8+ Kh7 Ree3 Qd1+ Re1 Qh5 fxg7 Rxg7 Rg2 Qf3 Rf1
Qe3+ Rff2 Kg8 Rg4 Qe6 Rg5 Qe3 Rg2 Kh7 a3 Qd4 Qxd4
cxd4 Kf1 Re7 Rg4 c5
27 -1.41 00:17 126948K d3 Qc3 Rf7 Rg3 Rf8 Rf1 Rf7 Rff3 Qe2 Re3 Qd1+ Re1 Qh5
Re8+ Kh7 Ree3 Qd1+ Re1 Qh5 fxg7 Rxg7 Rg2 Qf3 Rf1
Qe3+ Rff2 Kg8 Rg4 Qe6 Rg5 Qe3 Rg2 Kh7 a3 Qh6 Rg3
28 +0.00 00:24 184198K Rb3 axb3 Qe1+ Kg2 Be4+ Kg3 Qe3+ Kg4 Qxf2 Rxg7+ Kh8
Qxc5 Qf3+ Kg5 Qg3+ Kh5 Qh3+ Kg5 Qg3+
29 +0.00 00:26 199620K Rb3 axb3 Qe1+ Kg2 Be4+ Kg3 Qe3+ Kg4 Qxf2 Rxg7+ Kh8
Qxc5 Qf3+ Kg5 Qg3+ Kh5 Qh3+ Kg5 Qg3+
30 +0.00 00:27 210924K Rb3 axb3 Qe1+ Kg2 Be4+ Kg3 Qe3+ Kg4 Qxf2 Rxg7+ Kh8
Qxc5 Qf3+ Kg5 Qg3+ Kh5 Qh3+ Kg5 Qg3+
31 +0.00 00:31 244454K Rb3 axb3 Qe1+ Kg2 Be4+ Kg3 Qe3+ Kg4 Qxf2 Rxg7+ Kh8
Qxc5 Qf3+ Kg5 Qg3+ Kh5 Qh3+ Kg5 Qg3+
Nodes: 244851105
Nodes/second: 7751887
Best move: Rb3
Ponder move: axb3
Code: Select all
Critter v1.6a 64-bit, by Richard Vida
hardware POPCNT enabled
setboard 6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
info string Opening book: "book.cbk" - 57853 entries [native]
info string GTB Init OK (5 piece set)
info string Using 8 thread(s), 2048 MB hash LP=off
setoption name Hash value 4096
go infinite
info string Using 8 thread(s), 4096 MB hash LP=off
2/ 4 00:00 4113 228500 -0.63 Qe1+ Rf1 Qe8 Rxc5 gxf6 Rxf6
2/ 5 00:00 4316 239777 -0.38 gxf6 Rxf6 Qe1+ Rf1
3/ 7 00:00 5034 239714 -0.57 gxf6 Rxf6 Qe1+ Rf1 Qe8
4/ 8 00:00 5747 261227 -0.55 gxf6 Rxf6 Qe1+ Rf1 Qe8 Qxc5 Rxb2 Qxd4 Rxa2
5/ 9 00:00 8575 276612 -0.58 gxf6 Rxf6 Qe1+ Rf1 Qe4 Rg2 Rb4
6/11 00:00 15609 283800 -0.64 gxf6 Rxf6 Qe1+ Rf1 Qe4 Rg2 Rb4 Qg3
7/16 00:00 45994 277072 -0.58 gxf6 Rxf6 Rg7 Qh3 Qe1+ Rf1 Qe2 Qg3 Kh7 Rxc5 Qxb2 Rxc6 Qxa2
7/18 00:00 132025 278533 -0.33 d3 Qc3 a3 b3 Rf7 Rg3 gxf6 Rxf6 Rxf6 Qxf6 Qe1+ Kg2 Qe2+ Qf2
8/21+ 00:00 172564 309809 -0.25 d3 Qc3 a3 b3 Rf7 Rg3 gxf6 Rxf6 Rxf6 Qxf6 Qe1+ Kg2 Qe2+ Qf2
8/23- 00:00 221278 315210 -0.42 d3 Qc3 a3 b3 Rf7 Rg3 gxf6 Rxf6 Rxf6 Qxf6 Qe1+ Kg2 Qe2+ Qf2
8/23 00:00 223950 316760 -0.26 d3 Qc3 gxf6 Qxf6 Qe1+ Kg2 Rg7 Qf4 Be4+ Kh3 Qh1+ Kg3 Qg1+ Kh4 Qh1+ Kg4 Qg1+ Kh5
9/23- 00:00 289497 341387 -0.34 d3 Qc3 gxf6 Qxf6 Qe1+ Kg2 Rg7 Qf4 Be4+ Kh3 Qh1+ Kg3 Qg1+ Kh4 Qh1+ Kg4 Qg1+ Kh5
9/23 00:00 307995 353610 -0.30 d3 Qc3 gxf6 Qxf6 Qe1+ Kg2 Rg7 Qf4 Be4+ Kg3 Qg1+ Kh4 Qh1+ Rh2 Qe1+ Kh3 Bg6 Qb8+ Qe8 Qxe8+ Bxe8 Rxc5
10/24- 00:01 468919 427845 -0.39 d3 Qc3 gxf6 Qxf6 Qe1+ Kg2 Rg7 Qf4 Be4+ Kg3 Qg1+ Kh4 Qh1+ Rh2 Qe1+ Kh3 Bg6 Qb8+ Qe8 Qxe8+ Bxe8 Rxc5
10/33+ 00:01 1004K 590957 -0.22 Rb3 axb3 gxf6 Rxc5 Qe1+ Kg2 Be4+ Kg3 Qg1+ Kf4 Qxf2+ Kxe4 Qe3+
10/39+ 00:01 1092K 608760 -0.09 Rb3 axb3 gxf6 Rxc5 Qe1+ Kg2 Be4+ Kg3 Qg1+ Kf4 Qxf2+ Kxe4 Qe3+
10/41 00:01 1201K 639239 +0.00 Rb3 f7+ Bxf7 axb3 Qe1+ Kg2 Qe4+ Rf3 Qe2+ Kg3 Qe1+ Kf4 Qc1+ Kf5 Qc2+ Ke5 Qe2+ Kf4 Qd2+ Kf5 Qc2+
-
- Posts: 910
- Joined: Wed Mar 08, 2006 9:46 pm
- Location: Plovdiv, Bulgaria
Re: 18 days from SF4 release and about ~30+ ELO gain!
Marco, you sorry? Why? I can't understand... You're sharing all your ideas (if not code) with all commercial authors, right? That's more than enough for not feeling any remorse for that what you're doing.mcostalba wrote:Improvement cannot be foreseen in advance. It could be that next month we got 0 ELO, it happens and you know it.Houdini wrote:Indeed, the pace of Stockfish improvement is amazing, the development framework constructed by Gary is awesome.
Clearly no individual or two-person team can keep up with this in the long run, so this could mean the end of commercial chess engines as we currently know them. Maybe in 2 years time only Stockfish and derivates will continue to be developed.
Robert
But there is another side effect of open development that could be more threatening for commercial engines, a threat that was not foreseen in advance and that even I didn't realize it would be a problem. This is the obsolescence of release process: just few days after Stockfish 4 is out, almost all power users have dismissed it in favor of last nightly build (I just come here now from Playchess where there is even not one SF 4 but are all nightly builds): this is something commercial engines have no defense against, simply they cannot do this. As long as the ELO gap is big it is ok, but when the open developed engine reaches the level of commercials, a new compile each day can really badly affect the commercial release because it greatly speeds up its obsolescent.
I have to say that this was not foreseen and I am sorry for this, it is not out target (daily binaries are even built outside of SF team) but it is something that, considered the open nature of the development, it is almost impossible to avoid.
Just my 2 cents.
take it easy
-
- Posts: 1600
- Joined: Mon Feb 21, 2011 9:48 am
Re: 18 days from SF4 release and about ~30+ ELO gain!
old SFS.Taylor wrote: nTCEC will now be using the old SF
(just kidin)
Two, three weeks.
Come on, do not make me laugh.
The rules are clear and equal for all.
And there is much difference, no.[/b]
-
- Posts: 284
- Joined: Tue Aug 13, 2013 9:44 am
Re: 18 days from SF4 release and about ~30+ ELO gain!
+1GenoM wrote:Marco, you sorry? Why? I can't understand... You're sharing all your ideas (if not code) with all commercial authors, right? That's more than enough for not feeling any remorse for that what you're doing.mcostalba wrote:Improvement cannot be foreseen in advance. It could be that next month we got 0 ELO, it happens and you know it.Houdini wrote:Indeed, the pace of Stockfish improvement is amazing, the development framework constructed by Gary is awesome.
Clearly no individual or two-person team can keep up with this in the long run, so this could mean the end of commercial chess engines as we currently know them. Maybe in 2 years time only Stockfish and derivates will continue to be developed.
Robert
But there is another side effect of open development that could be more threatening for commercial engines, a threat that was not foreseen in advance and that even I didn't realize it would be a problem. This is the obsolescence of release process: just few days after Stockfish 4 is out, almost all power users have dismissed it in favor of last nightly build (I just come here now from Playchess where there is even not one SF 4 but are all nightly builds): this is something commercial engines have no defense against, simply they cannot do this. As long as the ELO gap is big it is ok, but when the open developed engine reaches the level of commercials, a new compile each day can really badly affect the commercial release because it greatly speeds up its obsolescent.
I have to say that this was not foreseen and I am sorry for this, it is not out target (daily binaries are even built outside of SF team) but it is something that, considered the open nature of the development, it is almost impossible to avoid.
Just my 2 cents.
This thread has become ridiculous. Robert Houdart and others that followed almost made me shed tears.
-
- Posts: 6401
- Joined: Thu Mar 09, 2006 8:30 pm
- Location: Chicago, Illinois, USA
Re: 18 days from SF4 release and about ~30+ ELO gain!
What is a "dangerous" move?Uri Blass wrote:It seems to me that most of the progress that stockfish make is thanks to search improvements.
I feel that stockfish's search is still weak part of it otherwise stockfish's developement version could solve the following position fast
[D]6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98
I can accept being twice slower than most top programs but stockfish developement version seems to me more than 10 times slower than most top programs after the stockfish team decided to reduce dangerous moves because unfortunately it gave some elo improvement.
Note that my last patch that passed both stage I and stage II correct the problem but marco is going to test a simpler patch that helps to find Rb3 faster than the developement version but still in the order of 10 times slower relative to my patch(and it is not that my patch could help stockfish to find Rb3 faster than the best top programs so I do not feel that I extend too much).
Note that my intuition tells me that if we want stockfish to be strong against weak opponents then we need to have less positions when stockfish is significantly weaker than them in tactics.
Maybe my intuition is wrong but my intuition tells me that the patch of not reducing dangerous moves has to be counter productive
if we test against opponents that are 200 elo weaker that is important for rating lists.
Miguel
-
- Posts: 5106
- Joined: Tue Apr 29, 2008 4:27 pm
Re: 18 days from SF4 release and about ~30+ ELO gain!
Uri,Uri Blass wrote:
Note that my last patch that passed both stage I and stage II correct the problem but marco is going to test a simpler patch that helps to find Rb3 faster than the developement version but still in the order of 10 times slower relative to my patch(and it is not that my patch could help stockfish to find Rb3 faster than the best top programs so I do not feel that I extend too much).
If your patch is good you can make it and distribute it and you don't need anyone's permission because that is granted by the licence. But it's still not your program no matter what you do. Anyone who works on GPL is working on software that is owned by someone else. It works great if you can live with that and it's a great way to collaborate on software as long as you understand it's not yours and that you are providing free labor for a common interest under someone else's control.
Although I cannot see this ever happening they could even in the future "close the sources" and keep the changes contributed and there is nothing you could do. They OWN the licence. They could not retroactively close already released sources of course but they would continue to hold the licence for those older sources. They are the ONLY ones who could close the sources and continue to work and update them.
For this project the GPL is mostly not even relevant unless you plan to fork. If you work on commercial software for a company and you are one of 100 different employees in a coding "sweat shop" it's no different, you are working for them, they own the software copyright and they own the software. It's a good arrangement because they pay you for doing it and/or you love doing it anyway and you accept that and it's even better if they pat you on the back for being a good employee.
I have in the past considered opening up software - but it's complicated. You cannot even by law FORCE your software to remain truly open and free - because you must maintain ownership to do that - and then it's restricted again. When a version of Robbolito came out with the missing licence it was immediately grabbed and the source closed and now we have Houdini. If something is free to do as you wish - it means you can claim it as your own - but then it's no longer free for others. So it seems that someone must always maintain legal ownership unless you don't care if that happens.
So the downside is that Stockfish is not free and even though it's a community project it doesn't belong to the community. Otherwise, why isn't your name listed when you type "uci"?
Here is a conundrum. Remember when Robbolito was released without a licence? What would have happened if someone had immediately slapped a GPL 2 licence on it? Would they suddenly become the new owner and licence holder? Could they have put their own name in the credits?
I think your intuition is wrong. All pruning is invalid by your argument because you can cherry pick many positions where the wrong move is pruned.
Note that my intuition tells me that if we want stockfish to be strong against weak opponents then we need to have less positions when stockfish is significantly weaker than them in tactics.
Maybe my intuition is wrong but my intuition tells me that the patch of not reducing dangerous moves has to be counter productive
if we test against opponents that are 200 elo weaker that is important for rating lists.
Komodo has occasionally been criticized for being tactically weak because we have taken favorably ELO trade-offs. But the key words here are "favorable" and "trade-off" and if you take a more relevant view what matters is what the program can actually see. If Stockfish is stronger due to some patch but appears to score less on a tactical problem set, it still means it's seeing more of what is relevant. It's not pragmatic to trade seeing more for seeing less just because it helps a small subset of positions. Of course some people want that - so for example Houdini has a tactical mode for people such as yourself who don't mind having a weaker program if it's more tactical.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.