TCEC Season 8


jts

Recommended Posts

TCEC is Thoresen Chess Engines Competition. It is not exactly the official world computer chess championship but everybody knows it has more credibility than the official WCCC.

Season 8 begins August 21, 2015. Complete info. TCEC facebook page.

Currently the champion is Komodo. In the superfinal in Season 7, Komodo and Stockfish had a 64 game match. They were very evenly matched. There were lots of draws. It was not obvious until near the end who would win.

Some differences between Komodo and Stockfish:

1. Komodo is proprietary; Stockfish is open source.

2. Komodo is the most accurate evaluator with the help of GM Larry Kaufman; Stockfish is the deepest calculator.

Both Komodo and Stockfish have been updated. The other engines, I looked but didn't find info that the stronger ones such as Houdini and Gull were updated. From this page it seems that the latest version of Stockfish is stronger than the latest version of Komodo. So unless there is a last minute change or the time control makes a difference or the number of cores makes a difference or the RAM makes a difference, I think probably Stockfish will beat Komodo in the superfinal and again become the champion.

Quoting the article on chessdom.com: "The TCEC Season 8 will start on a 20 core Xeon server with 128 GB of RAM. The final stages will be played on a 40 cores server."

In the TCEC Season 8 Superfinal, we will see the highest quality chess ever seen on a chessboard in the entire history of chess. Until next time.

This is where it will happen.

Link to comment
Share on other sites

So both programs will run on the same computer? Sorry I'm new to this. Alsohelp me to understand the difference between "evaluator" vs "calculator"

Link to comment
Share on other sites

So both programs will run on the same computer? Sorry I'm new to this. Alsohelp me to understand the difference between "evaluator" vs "calculator"

The strength of a chess engine depends in part on the computer it is run on. More powerful computer, stronger chess. If you want to know which chess engine is the best, you need to run them all on the same computer. If you have one running on a microcomputer and the other running on a supercomputer, that's no fair.

How can you be an experienced chess player and not know the difference between evaluation and calculation? Evaluation means who is better, White or Black? And how much? The crudest possible precision of evaluation is: win, lose, draw. These engines evaluate positions to a precision of a hundredth of a pawn. A centipawn. That's precision, not to be confused with accuracy. Komodo evaluates better than any other engine. Calculation means the search tree. Stockfish calculates deeper than any other engine.

I know what you are about to say. Evaluation changes with depth of search. True. Komodo evaluates better given the same depth of search.

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate? I remember when I was just out of high school and playing someone of equal ability, we drew most games, and each of us won a few. So I read a couple of those little paperbacks on playing chess and he IMMEDIATELY knew I was cheating . . . or I became a different person overnight. With an average time limit for each play, the fastest computer will win. So, if a new Intel chip is put in the fastest should always win. Would a computer eventually think, I think therefore I am, or as Rand said, I am, therefore I think? Recently Stephen Hawking cautioned the cyberneticists to not make a self aware computer/robot. Will Jeopardy have another tournament of human vs. computer, or might they just have three computers competing?

Link to comment
Share on other sites

Until computers can reproduce on their own they can be as "sentient" as they want. Until then we have the luxury of worrying about people, not computer sentience--just like the "sentient" computer. ("Please--don't pull that plug!")

--Brant

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate? I remember when I was just out of high school and playing someone of equal ability, we drew most games, and each of us won a few. So I read a couple of those little paperbacks on playing chess and he IMMEDIATELY knew I was cheating . . . or I became a different person overnight. With an average time limit for each play, the fastest computer will win. So, if a new Intel chip is put in the fastest should always win. Would a computer eventually think, I think therefore I am, or as Rand said, I am, therefore I think? Recently Stephen Hawking cautioned the cyberneticists to not make a self aware computer/robot. Will Jeopardy have another tournament of human vs. computer, or might they just have three computers competing?

Stalemate is not the only way a game of chess can be a draw. A draw can be: by agreement, by drawn endgame, by repetition, by stalemate, by 50 move rule.

Two programs of equal strength will on average win and lose the same number of games. The result will not be the same every game.

Doubling the speed of the computer will increase the Elo rating points; but will not result in a win every time against normal speed.

Computers will not achieve consciousness. Computers are machines that follow instructions. Computers do only what they are told to do. Computers on their own, without programmers to 'breathe the breath of life into them' (figuratively speaking), are as stupid as a box of rocks.

Link to comment
Share on other sites

I know what you are about to say. Evaluation changes with depth of search.

this is what I was going to say. It would seem that humans are/were better evaluators then computers who had speed and depth on their side, but that speed and depth gives a good proxy for evaluation. In fact it was/is good enough to win

Link to comment
Share on other sites

Until computers can reproduce on their own they can be as "sentient" as they want. Until then we have the luxury of worrying about people, not computer sentience--just like the "sentient" computer. ("Please--don't pull that plug!")

--Brant

I used to worry about the SkyNet problem, but then I thought it through logically and I agree with you. If there is anything to fear it is the people behind the machines, not the machines themselves--even sentient machines.

Link to comment
Share on other sites

Brant wrote: Until computers can reproduce on their own they can be as "sentient" as they want. Until then we have the luxury of worrying about people, not computer sentience--just like the "sentient" computer. ("Please--don't pull that plug!")
end quote

So, like drones and water balloons we need only fear teenagers and psychopaths, who have access to more and more powerful computers, and robots.

Link to comment
Share on other sites

Was it Spielberg’s movie, Minority Report with Tom Cruise that had horseshoe crab-sized robots that moved like spiders? We are close to that level of technology now with beetle sized robots with cameras that can creep into a building. If it’s battery only lasts an hour that is its greatest limitation.

Think of this: four helicopter type drones with cameras doing shifts of near silent and invisible surveillance. Perhaps three operators remotely controlling them along with onboard computers, 24 hours a day. For weeks. What do drones with movie cameras cost now? Less than 500 bucks each? How much would three ex-army guys cost per day to monitor them 24/7? Sam Spade, Spencer for Hire, and Mike Hammer types will be using those systems to find out where the cheating spouse is headed each day. And the applications for spying on ambassadors is great. I read where a lot of clandestine meetings are taking place in parking garages to avoid surveillance.

And the next time you see someone using their cell phone to take a picture of their walking companion, they are really tailing you and taking your picture. Smile!

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate? I remember when I was just out of high school and playing someone of equal ability, we drew most games, and each of us won a few. So I read a couple of those little paperbacks on playing chess and he IMMEDIATELY knew I was cheating . . . or I became a different person overnight. With an average time limit for each play, the fastest computer will win. So, if a new Intel chip is put in the fastest should always win. Would a computer eventually think, I think therefore I am, or as Rand said, I am, therefore I think? Recently Stephen Hawking cautioned the cyberneticists to not make a self aware computer/robot. Will Jeopardy have another tournament of human vs. computer, or might they just have three computers competing?

Stalemate is not the only way a game of chess can be a draw. A draw can be: by agreement, by drawn endgame, by repetition, by stalemate, by 50 move rule.

Two programs of equal strength will on average win and lose the same number of games. The result will not be the same every game.

Doubling the speed of the computer will increase the Elo rating points; but will not result in a win every time against normal speed.

Computers will not achieve consciousness. Computers are machines that follow instructions. Computers do only what they are told to do. Computers on their own, without programmers to 'breathe the breath of life into them' (figuratively speaking), are as stupid as a box of rocks.

Computers are--yeah, they are. We do not know how they will evolve but we do know they will be continuously evolved. Maybe some genius will throw some biology into them. Or computers into people. All our machines are extensions of ourselves and computers right now are existential mental tools. When they become intrinsic--if they do--then they'll achieve consciousness through parasitical human brain modification. One's thoughts will be the computer's thoughts.

This might create a problem if the computer spots a pretty laptop.

--Brant

a problem for the future

Link to comment
Share on other sites

Brant wrote: When they become intrinsic--if they do--then they'll achieve consciousness through parasitical human brain modification. One's thoughts will be the computer's thoughts.

end quote

That does seem to follow the logical road. We ARE enhancing our brains with machines now. Prosthetic eyes are a reality though the last I heard they only enabled a former blind person to *identify* light and dark and large objects. But for a blind person that is a huge advantage. Hearing aids and implants like Rush Limbaugh owns are a reality. Attached phones. Night vision glasses. Human battle bots like in Avatar. We may be close to Borg like humans who can plug in enhancing devices. And I think direct computer enhanced thinking is coming. Having search engines on your computer is like having a super, always updating, encyclopedia.

I have visited two grave yards recently to change flowers and clean up. I noticed devices on graves that played pictures and sounds. Some graves have those glowing solar lights. And what about enhanced remembering devices beyond a photo album? How about an interacting memorial? A family friend who died still has an unopened can of Coors on his grave, though humans have put mementoes on graves since we evolved. Robotic pets that mimic poor Scout who was run over? Sigh.

We are also on a path to machine intelligence though many think it will just seem to be intelligent but it is only fulfilling its programming. We have satellites with primitive machine intelligence zooming past Pluto and heading out of our solar system. Are they our non-human ambassadors to the stars?

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate?

Actually, white should win every game.

A...

Link to comment
Share on other sites

Until computers can reproduce on their own they can be as "sentient" as they want. Until then we have the luxury of worrying about people, not computer sentience--just like the "sentient" computer. ("Please--don't pull that plug!")

--Brant

You mean as sentient as some human wants??????

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate?

Actually, white should win every game.

A...

Some interesting statistics from TCEC Season 7 Superfinal.

Komodo and Stockfish had 64 battle royales between them. Most of the games were a draw. There were 11 decisive games, non-draw games. One very interesting fact is that all 11 decisive games were won by White. Not one win by Black. Komodo won 6; Stockfish won 5.

Another interesting fact. The first 8 moves on each side were pre-decided. There were 32 openings, pre-decided, each played twice with colors reversed. The interesting fact is some openings White won both games.

In one world title match between the 2 super Ks, Karpov vs Kasparov, as usual most of the games were a draw, 9 were decisive, White won 8, Black won 1.

White starts with the unstable advantage of first move which gives White an initiative. For the first few moves, White is hammer, Black is anvil; White plays the fiddle, Black dances to White's tune. Bobby Fischer said, "With Black you've got to equalize first before looking for something." To equalize means to neutralize White's initiative.

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate?

Actually, white should win every game.

A...

Some interesting statistics from TCEC Season 7 Superfinal.

Komodo and Stockfish had 64 battle royales between them. Most of the games were a draw. There were 11 decisive games, non-draw games. One very interesting fact is that all 11 decisive games were won by White. Not one win by Black. Komodo won 6; Stockfish won 5.

Another interesting fact. The first 8 moves on each side were pre-decided. There were 32 openings, pre-decided, each played twice with colors reversed. The interesting fact is some openings White won both games.

In one world title match between the 2 super Ks, Karpov vs Kasparov, as usual most of the games were a draw, 9 were decisive, White won 8, Black won 1.

White starts with the unstable advantage of first move which gives White an initiative. For the first few moves, White is hammer, Black is anvil; White plays the fiddle, Black dances to White's tune. Bobby Fischer said, "With Black you've got to equalize first before looking for something." To equalize means to neutralize White's initiative.

Correct.

Whenever I play someone for the first time, I prefer to play black because my pawn structures are almost always superior than most good players and occasionally superior to excellent players.

Loving me my Hans Kmoch's Pawn Power which is the best structural book on chess that revolutionized my game.

A...

Link to comment
Share on other sites

  • 3 weeks later...

Both Komodo and Stockfish have been updated. The other engines, I looked but didn't find info that the stronger ones such as Houdini and Gull were updated. From this page it seems that the latest version of Stockfish is stronger than the latest version of Komodo. So unless there is a last minute change or the time control makes a difference or the number of cores makes a difference or the RAM makes a difference, I think probably Stockfish will beat Komodo in the superfinal and again become the champion.

[.......]

This is where it will happen.

In stage 1b, Stockfish won all 11 games. True it got a lucky break a time or 2 when its opponent stalled. But I think Stockfish would have won those games anyway. The game Houdini - Stockfish was an epic battle that did credit to both sides. "The stern joy that warriors feel in foemen worthy of their steel". Much entertainment for the commentators. The mighty fish demonstrated to Houdini who the real magician is.

In stage 2, 3 engines are disqualified because they stalled too many times.

In stage 2, the mighty fish and the terrible lizard (Komodo) will meet formally for the first time (latest version of each). The lizard was the strongest in stage 1a.

Link to comment
Share on other sites

  • 1 month later...

Stage 3 is on its last round and next is the strongest chess match in history between Komodo and Stockfish. See it here.

It will be a clash of titans and the earth will shake.

Both engines are Elo 3200+, more than 300 Elo rating points stronger than the strongest human.

In stage 3, Stockfish developed a time management bug and lost 1.5 points by overstepping the time limit before they replaced it with an earlier version. Then it lost another 0.5 point on time. Stockfish qualified for the superfinal in spite of losing 2.0 points to this bug. I hope the Stockfish programmers fix this bug before the superfinal.

In a blitz match, 1 core vs 1 core, Stockfish destroyed Komodo. But it seems at normal time control, 20 cores vs 20 cores, they are more evenly matched.

Link to comment
Share on other sites

If two computer playing chess programs have the same power will they always stalemate?

Actually, white should win every game.

A...

Has that been proven??? Why couldn't black force a statemate for every possible white opening. Proving white has a sure win would be a rather difficult thing to prove mathematically in the case of chess. On other games like variants of tic-tac-toe (Go Moku is the generalized version) is can be shown the player moving second can always force a draw for any size Go Moku board.

Now here is a twist on the problem Suppose there were a computer and memory combination large enough to test any chess game permissible under FIDE rules. And suppose the computer program shows that for some white openings white has a sure win and for other white openings white has a forced loss. How would one prove that the program running that "proof" did not have a logic error in it that violated some FIDE rule? Proving the program is sound may be just as hard as proving something about the game.

Link to comment
Share on other sites

Actually, white should win every game.

A...

Has that been proven??? Why couldn't black force a statemate for every possible white opening. Proving white has a sure win would be a rather difficult thing to prove mathematically in the case of chess. On other games like variants of tic-tac-toe (Go Moku is the generalized version) is can be the player moving second can always force a draw for any size Go Moku board.

All I can say Bob, and I do not have the math to prove it, however, every time I have played in a tournament, as an individual, or, on a team, white win got 2 points, Stalemates got 1 pt.

If I could get a draw as black, that was a win. Kinda like breaking the opponents serve in tennis.

Intercepting a pass in football and returning it for a touchdown is similar in magnitude.

A...

Link to comment
Share on other sites

Actually, white should win every game.

A...

Has that been proven??? Why couldn't black force a statemate for every possible white opening. Proving white has a sure win would be a rather difficult thing to prove mathematically in the case of chess. On other games like variants of tic-tac-toe (Go Moku is the generalized version) is can be the player moving second can always force a draw for any size Go Moku board.

All I can say Bob, and I do not have the math to prove it, however, every time I have played in a tournament, as an individual, or, on a team, white win got 2 points, Stalemates got 1 pt.

If I could get a draw as black, that was a win. Kinda like breaking the opponents serve in tennis.

Intercepting a pass in football and returning it for a touchdown is similar in magnitude.

A...

How many of the combinatorially possible Chess games have you played? There are more possible Chess Games than there are atoms on earth.

Link to comment
Share on other sites

Actually, white should win every game.

A...

Has that been proven??? Why couldn't black force a statemate for every possible white opening. Proving white has a sure win would be a rather difficult thing to prove mathematically in the case of chess. On other games like variants of tic-tac-toe (Go Moku is the generalized version) is can be the player moving second can always force a draw for any size Go Moku board.

All I can say Bob, and I do not have the math to prove it, however, every time I have played in a tournament, as an individual, or, on a team, white win got 2 points, Stalemates got 1 pt.

If I could get a draw as black, that was a win. Kinda like breaking the opponents serve in tennis.

Intercepting a pass in football and returning it for a touchdown is similar in magnitude.

A...

How many of the combinatorially possible Chess games have you played? There are more possible Chess Games than there are atoms on earth.

Yes Bob, I understand that.

What is your point in making that statement about stalemates?

Link to comment
Share on other sites

All I can say Bob, and I do not have the math to prove it, however, every time I have played in a tournament, as an individual, or, on a team, white win got 2 points, Stalemates got 1 pt.

If I could get a draw as black, that was a win. Kinda like breaking the opponents serve in tennis.

Intercepting a pass in football and returning it for a touchdown is similar in magnitude.

A...

You say a white win got 2 points. How many points did a black win get? How many points did a non-stalemate draw get?

The normal rules are a win by either color gets 1 point, a loss gets 0, a draw gets half a point each.

One nonstandard kind of scoring to solve the problem of too many draws is 3 points for a win, 0 for a loss, 1 point each for a draw.

Arno Nickel, former world correspondence chess champion, proposed that stalemates (being a special case of a draw) should be worth something more than a draw and less than a win. Read about his proposal here.

On the subject of whether white (with the advantage of first move) has a theoretical win: Not even the most powerful chess engines in 2015 can prove it one way or the other. The intuition of all grandmasters that have commented on the first move advantage that I know of says chess is a theoretical draw. Bobby Fischer said it's almost definite it's a draw.

Read all about the advantage of first move here.

Link to comment
Share on other sites

All I can say Bob, and I do not have the math to prove it, however, every time I have played in a tournament, as an individual, or, on a team, white win got 2 points, Stalemates got 1 pt.

If I could get a draw as black, that was a win. Kinda like breaking the opponents serve in tennis.

Intercepting a pass in football and returning it for a touchdown is similar in magnitude.

A...

You say a white win got 2 points. How many points did a black win get? How many points did a non-stalemate draw get?

The normal rules are a win by either color gets 1 point, a loss gets 0, a draw gets half a point each.

One nonstandard kind of scoring to solve the problem of too many draws is 3 points for a win, 0 for a loss, 1 point each for a draw.

Arno Nickel, former world correspondence chess champion, proposed that stalemates (being a special case of a draw) should be worth something more than a draw and less than a win. Read about his proposal here.

On the subject of whether white (with the advantage of first move) has a theoretical win: Not even the most powerful chess engines in 2015 can prove it one way or the other. The intuition of all grandmasters that have commented on the first move advantage that I know of says chess is a theoretical draw. Bobby Fischer said it's almost definite it's a draw.

Read all about the advantage of first move here.

Thanks a lot JTS.

There was no differentiation between an agreed to draw and a stalemate.

My recollection was that if you both made the same sequence of moves for three/four? turns, the rules of the league, or, the tournament.

Whichever side won got +2 and each side got 1 point for any draw[agreed], actual stalemate, and the same consecutive moves stalemate.

A...

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now