ICGA's 2015 World Computer Chess Championship/Events

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
mhull
Posts: 12263
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by mhull » Mon Jul 06, 2015 3:36 pm

Laskos wrote:
lucasart wrote:
Laskos wrote:
lucasart wrote:even without book/tb and running on a cell phone, sf (or komodo) could win this pathetic event…
Well, 24 i7 cores, book, TBs were not sufficient.
Yes, this result is quite surprising. Despite Jonny's monstruous CPU horse-power, I still expected Komodo to win.

But comparing engines, using only 8 games, with different hardware, different opening books, and manual operation (which causes extra lags that that can cause engines to lose on time)... As a statistician, wouldn't you say that's really stupid?
I am am not saying WCCC is of some worth, at least for someone who needs the strongest engine. Just a funny competition combining the dice with some skills (not only the playing skills in same conditions, there are hardware skills, book skills), more like a soccer match or even more random. You replied with the cell phone SF to win WCCC to a discussion initiated by my post, that's why I remembered the topic:
Laskos wrote:If there are 11 engines, 10 games each engine, and say 5 engines 200 Elo points weaker than SF at LTC and equal hardware, 4 engines 400 Elo weaker, then taking into account the hardware fluctuations and book fluctuations, SF has LESS than 50% to win WCCC.
WCCC is of some worth. It shows that lobotomized "tournaments" like CCRL are not an indication of how an un-lobotomized tournament (i.e. the AI has all it's AI turned on, position learning, book learning, pondering, extended resource management, etc.) will end.
Matthew Hull

User avatar
Laskos
Posts: 9244
Joined: Wed Jul 26, 2006 8:21 pm
Full name: Kai Laskos

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by Laskos » Mon Jul 06, 2015 4:53 pm

mhull wrote:
Laskos wrote:
lucasart wrote:
Laskos wrote:
lucasart wrote:even without book/tb and running on a cell phone, sf (or komodo) could win this pathetic event…
Well, 24 i7 cores, book, TBs were not sufficient.
Yes, this result is quite surprising. Despite Jonny's monstruous CPU horse-power, I still expected Komodo to win.

But comparing engines, using only 8 games, with different hardware, different opening books, and manual operation (which causes extra lags that that can cause engines to lose on time)... As a statistician, wouldn't you say that's really stupid?
I am am not saying WCCC is of some worth, at least for someone who needs the strongest engine. Just a funny competition combining the dice with some skills (not only the playing skills in same conditions, there are hardware skills, book skills), more like a soccer match or even more random. You replied with the cell phone SF to win WCCC to a discussion initiated by my post, that's why I remembered the topic:
Laskos wrote:If there are 11 engines, 10 games each engine, and say 5 engines 200 Elo points weaker than SF at LTC and equal hardware, 4 engines 400 Elo weaker, then taking into account the hardware fluctuations and book fluctuations, SF has LESS than 50% to win WCCC.
WCCC is of some worth. It shows that lobotomized "tournaments" like CCRL are not an indication of how an un-lobotomized tournament (i.e. the AI has all it's AI turned on, position learning, book learning, pondering, extended resource management, etc.) will end.
Combined with how a perfect dice works.

User avatar
mhull
Posts: 12263
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by mhull » Mon Jul 06, 2015 6:24 pm

Laskos wrote:
mhull wrote:
Laskos wrote:
lucasart wrote:
Laskos wrote:
lucasart wrote:even without book/tb and running on a cell phone, sf (or komodo) could win this pathetic event…
Well, 24 i7 cores, book, TBs were not sufficient.
Yes, this result is quite surprising. Despite Jonny's monstruous CPU horse-power, I still expected Komodo to win.

But comparing engines, using only 8 games, with different hardware, different opening books, and manual operation (which causes extra lags that that can cause engines to lose on time)... As a statistician, wouldn't you say that's really stupid?
I am am not saying WCCC is of some worth, at least for someone who needs the strongest engine. Just a funny competition combining the dice with some skills (not only the playing skills in same conditions, there are hardware skills, book skills), more like a soccer match or even more random. You replied with the cell phone SF to win WCCC to a discussion initiated by my post, that's why I remembered the topic:
Laskos wrote:If there are 11 engines, 10 games each engine, and say 5 engines 200 Elo points weaker than SF at LTC and equal hardware, 4 engines 400 Elo weaker, then taking into account the hardware fluctuations and book fluctuations, SF has LESS than 50% to win WCCC.
WCCC is of some worth. It shows that lobotomized "tournaments" like CCRL are not an indication of how an un-lobotomized tournament (i.e. the AI has all it's AI turned on, position learning, book learning, pondering, extended resource management, etc.) will end.
Combined with how a perfect dice works.
Un-lobotomizing chess AI is the only way a computer chess tournament has any chance of being as interesting and fun as a human chess tournament.
Matthew Hull

User avatar
Graham Banks
Posts: 32779
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by Graham Banks » Mon Jul 06, 2015 7:28 pm

mhull wrote:
Laskos wrote:
mhull wrote:
Laskos wrote:
lucasart wrote:
Laskos wrote:
lucasart wrote:even without book/tb and running on a cell phone, sf (or komodo) could win this pathetic event…
Well, 24 i7 cores, book, TBs were not sufficient.
Yes, this result is quite surprising. Despite Jonny's monstruous CPU horse-power, I still expected Komodo to win.

But comparing engines, using only 8 games, with different hardware, different opening books, and manual operation (which causes extra lags that that can cause engines to lose on time)... As a statistician, wouldn't you say that's really stupid?
I am am not saying WCCC is of some worth, at least for someone who needs the strongest engine. Just a funny competition combining the dice with some skills (not only the playing skills in same conditions, there are hardware skills, book skills), more like a soccer match or even more random. You replied with the cell phone SF to win WCCC to a discussion initiated by my post, that's why I remembered the topic:
Laskos wrote:If there are 11 engines, 10 games each engine, and say 5 engines 200 Elo points weaker than SF at LTC and equal hardware, 4 engines 400 Elo weaker, then taking into account the hardware fluctuations and book fluctuations, SF has LESS than 50% to win WCCC.
WCCC is of some worth. It shows that lobotomized "tournaments" like CCRL are not an indication of how an un-lobotomized tournament (i.e. the AI has all it's AI turned on, position learning, book learning, pondering, extended resource management, etc.) will end.
Combined with how a perfect dice works.
Un-lobotomizing chess AI is the only way a computer chess tournament has any chance of being as interesting and fun as a human chess tournament.
I still find most of my tournaments interesting. Many of the engine authors do as well.
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

User avatar
mhull
Posts: 12263
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by mhull » Mon Jul 06, 2015 7:58 pm

Graham Banks wrote:
mhull wrote:
Laskos wrote:
mhull wrote:
Laskos wrote:
lucasart wrote:
Laskos wrote:
lucasart wrote:even without book/tb and running on a cell phone, sf (or komodo) could win this pathetic event…
Well, 24 i7 cores, book, TBs were not sufficient.
Yes, this result is quite surprising. Despite Jonny's monstruous CPU horse-power, I still expected Komodo to win.

But comparing engines, using only 8 games, with different hardware, different opening books, and manual operation (which causes extra lags that that can cause engines to lose on time)... As a statistician, wouldn't you say that's really stupid?
I am am not saying WCCC is of some worth, at least for someone who needs the strongest engine. Just a funny competition combining the dice with some skills (not only the playing skills in same conditions, there are hardware skills, book skills), more like a soccer match or even more random. You replied with the cell phone SF to win WCCC to a discussion initiated by my post, that's why I remembered the topic:
Laskos wrote:If there are 11 engines, 10 games each engine, and say 5 engines 200 Elo points weaker than SF at LTC and equal hardware, 4 engines 400 Elo weaker, then taking into account the hardware fluctuations and book fluctuations, SF has LESS than 50% to win WCCC.
WCCC is of some worth. It shows that lobotomized "tournaments" like CCRL are not an indication of how an un-lobotomized tournament (i.e. the AI has all it's AI turned on, position learning, book learning, pondering, extended resource management, etc.) will end.
Combined with how a perfect dice works.
Un-lobotomizing chess AI is the only way a computer chess tournament has any chance of being as interesting and fun as a human chess tournament.
I still find most of my tournaments interesting. Many of the engine authors do as well.
I've watched some of the CCRL games on the live viewer with some interest. However, it's not quite as compelling as watching WCCC games live (when that is even possible), or even CCT and ACCA games live on ICC (when that used to be possible -- not anymore without $$). Because the programs are running with books the authors would not choose for a competition, nor are the programs allowed to learn, one cannot consider the matches as valid tests of a program's true merits. We would never do that to a human player.

There is another way, IMHO.
Matthew Hull

User avatar
Laskos
Posts: 9244
Joined: Wed Jul 26, 2006 8:21 pm
Full name: Kai Laskos

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by Laskos » Mon Jul 06, 2015 9:49 pm

mhull wrote: Un-lobotomizing chess AI is the only way a computer chess tournament has any chance of being as interesting and fun as a human chess tournament.
The notion of fun is subjective. If the merits of programmer are important, then one would rather see WCSC instead of WCCC. Jonny winning WCCC isn't that illuminating in many things other than the hardware used. Although I agree that watching 8 engines round robin at LTC on custom hardware and books can bring more excitement purely on grounds of subjective excitement, very similar to that brought by human chess tourneys.

User avatar
mhull
Posts: 12263
Joined: Wed Mar 08, 2006 8:02 pm
Location: Dallas, Texas
Full name: Matthew Hull

Re: ICGA's 2015 World Computer Chess Championship/Events

Post by mhull » Tue Jul 07, 2015 3:16 am

Laskos wrote:
mhull wrote: Un-lobotomizing chess AI is the only way a computer chess tournament has any chance of being as interesting and fun as a human chess tournament.
The notion of fun is subjective. If the merits of programmer are important, then one would rather see WCSC instead of WCCC. Jonny winning WCCC isn't that illuminating in many things other than the hardware used. Although I agree that watching 8 engines round robin at LTC on custom hardware and books can bring more excitement purely on grounds of subjective excitement, very similar to that brought by human chess tourneys.
I would rather not see WCSC unless it was for very small uniform hardware footprints in the spirit of demoscene contests.

https://en.wikipedia.org/wiki/Demoscene

The interesting prospect for such a contest in chess might be the revival of a dedicated machine featuring the latest uniform platform champion and part of the prize, a contract for the winning programmer with the manufacturer.
Matthew Hull

Post Reply