The Real Kings of Chess Are Computers

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
User avatar
fourthirty
Full Member
Posts: 763
Joined: Fri Dec 06, 2013 8:46 pm
Location: San Francisco

The Real Kings of Chess Are Computers

Post by fourthirty »

Nice WSJ article on the TCEC competition:

http://www.wsj.com/articles/the-real-ki ... 1420827071
User avatar
Cyberchess
Full Member
Posts: 658
Joined: Wed Jan 08, 2014 6:10 pm

Post by Cyberchess »

:) Nice article!

Thanks for posting...

John
User avatar
pr1uk
Member
Posts: 220
Joined: Wed Apr 02, 2014 2:01 pm
Location: Strood, Kent. UK

Post by pr1uk »

A great read and once again thanks for posting



Peter
King Performance Chess Computer M830
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Post by Monsieur Plastique »

Well as the risk of being the party pooper, whilst I respect the engine hobbyists (especially their never ending dedication to performing engine / hardware comparisons together with an admirably pedantic approach that leaves most dedicated enthusiasts in the dust), I am never able to understand the attraction of two chess entities playing each other in games that are simply impossible for me (and I honestly suspect nearly all humans) to fully understand. If I can't understand what I am watching I don't see the point - I like to see a machine make a potential error and I want to understand that error and play the consequences of that error through in my mind in real time whilst the opponent is calculating it's own response. I still have happy and vivid memories of the computer matches my brother and I played in the 80s where all the excitement was down to seeing an error being committed and seeing if the computer opponent picked up on it or not.

With these modern 3300 ELO engines, the tactical implications in many positions are so incredibly deep that a human would require weeks or possibly months of studying just one tournament game in order to appreciate the full ramifications such is the sheer depth and width of calculation involved. I certainly don't believe that any human - even World Champions - can appreciate fully these games in real time and for me that takes nearly all the attraction out of it.

This is why I just love the old machines - I can play matches between two 1600 ELO machines, for example, and I have a chance of understanding the errors as they happen. And if I can't understand, it does not require a powerful engine to explain the errors to me. About the strongest chess playing entities I am interested in watching are machines like the Expert Travel and to a lesser extend my Fritz Nintendo, which are both machines around the low 1900 range or 2000 range depending on what rating scale you are subscribing to. After that, I am left bewildered. I even received a game from one of our members a couple of months back (Obsidian vs Chess Challenger) and it just left me completely flummoxed. I had absolutely no idea of what was going on in the game and had to sheepishly tell the other member that I was completely clueless (ironically, the only answer was to get an engine to analyse it).

Don't get me wrong - I have my fair share of engines, my main one being HIARCS on my main desktop, Fritz 13 on my laptop and Genius on my Smartphone. I am even contemplating purchasing Komodo simply because of the deep involvement of Larry Kaufmann.

This beings me to my last point. Given how ridiculously powerful processing power is these days, I would actually like to see a relatively weak engine (i.e about FIDE Master strength) that is so packed with enormous amounts of chess knowledge (far more than anything around currently) that the number of positions it can grind through is hardly any more than the 6502 machines of the 80s managed. Or even less.

I think we should now be turning our attention away from trying to break ELO barriers and actually creating significantly weaker machines and engines. I think these would be far more attractive opponents and for my own part, I'd get far more pride out of snatching a lucky draw against a 2200 ELO engine with hitherto unchartered depths of chess knowledge, then a program which is breaking ELO barriers.

Afterall, statistically most humans are never going to have a hope against even a 2200 ELO engine, so what is the point of losing to a 3300 ELO engine when you are going to get crushed by one 1100 points weaker? You might as well lose to the one that gives you the most human like game possible and which exhibits human strengths and weaknesses, without descending into the realm of actually making outright blunders. And for the really strong humans, well these engines have already well and truly surpassed them, so there are no humans around that can complain there isn't an engine strong enough to challenge them.
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
pr1uk
Member
Posts: 220
Joined: Wed Apr 02, 2014 2:01 pm
Location: Strood, Kent. UK

Post by pr1uk »

Monsieur Plastique i have to agree with you and you made a good argument for your views, but i can understand why people like to complete computer against computer or engines of course. As for engines i do have one on my iPad but i always transfer onto a chess board as to me chess is a 3D game. To me there is no better joy then bending over a real chess board and physically moving wooden pieces even better if i win now and again.




Peter
King Performance Chess Computer M830
User avatar
Fernando
Admiral of the Fleet
Posts: 3059
Joined: Tue Jul 31, 2007 4:35 pm
Location: Santiago de Chile

Post by Fernando »

Monsieur Plastique wrote:Well as the risk of being the party pooper, whilst I respect the engine hobbyists (especially their never ending dedication to performing engine / hardware comparisons together with an admirably pedantic approach that leaves most dedicated enthusiasts in the dust), I am never able to understand the attraction of two chess entities playing each other in games that are simply impossible for me (and I honestly suspect nearly all humans) to fully understand. If I can't understand what I am watching I don't see the point - I like to see a machine make a potential error and I want to understand that error and play the consequences of that error through in my mind in real time whilst the opponent is calculating it's own response. I still have happy and vivid memories of the computer matches my brother and I played in the 80s where all the excitement was down to seeing an error being committed and seeing if the computer opponent picked up on it or not.

With these modern 3300 ELO engines, the tactical implications in many positions are so incredibly deep that a human would require weeks or possibly months of studying just one tournament game in order to appreciate the full ramifications such is the sheer depth and width of calculation involved. I certainly don't believe that any human - even World Champions - can appreciate fully these games in real time and for me that takes nearly all the attraction out of it.

This is why I just love the old machines - I can play matches between two 1600 ELO machines, for example, and I have a chance of understanding the errors as they happen. And if I can't understand, it does not require a powerful engine to explain the errors to me. About the strongest chess playing entities I am interested in watching are machines like the Expert Travel and to a lesser extend my Fritz Nintendo, which are both machines around the low 1900 range or 2000 range depending on what rating scale you are subscribing to. After that, I am left bewildered. I even received a game from one of our members a couple of months back (Obsidian vs Chess Challenger) and it just left me completely flummoxed. I had absolutely no idea of what was going on in the game and had to sheepishly tell the other member that I was completely clueless (ironically, the only answer was to get an engine to analyse it).

Don't get me wrong - I have my fair share of engines, my main one being HIARCS on my main desktop, Fritz 13 on my laptop and Genius on my Smartphone. I am even contemplating purchasing Komodo simply because of the deep involvement of Larry Kaufmann.

This beings me to my last point. Given how ridiculously powerful processing power is these days, I would actually like to see a relatively weak engine (i.e about FIDE Master strength) that is so packed with enormous amounts of chess knowledge (far more than anything around currently) that the number of positions it can grind through is hardly any more than the 6502 machines of the 80s managed. Or even less.

I think we should now be turning our attention away from trying to break ELO barriers and actually creating significantly weaker machines and engines. I think these would be far more attractive opponents and for my own part, I'd get far more pride out of snatching a lucky draw against a 2200 ELO engine with hitherto unchartered depths of chess knowledge, then a program which is breaking ELO barriers.

Afterall, statistically most humans are never going to have a hope against even a 2200 ELO engine, so what is the point of losing to a 3300 ELO engine when you are going to get crushed by one 1100 points weaker? You might as well lose to the one that gives you the most human like game possible and which exhibits human strengths and weaknesses, without descending into the realm of actually making outright blunders. And for the really strong humans, well these engines have already well and truly surpassed them, so there are no humans around that can complain there isn't an engine strong enough to challenge them.
Agreed in everything. Just now deciding between playing Alexandra or Igor....

Fern
Festina Lente
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

Its for this reason I have little interest in the Human World Championships
anymore
long gone are the days where the winner can claim to be the strongest chess playing entity on earth
in fact...as the article clearly points out..the very best humans are playing several hundred elo and several classes below the best programs
all of which makes me ponder about the upside down nature of the prize funds..

the Human WC contestants wack up millions of dollars
while the winner(engine author ) of the Computer World Championship gets basically a Pizza and the beverage of his choice
OK... so its not that bad...sales of the winning program do sell nicely for a short time.. but its still a fraction of what the humans get for a 2 week match to play largely inferior chess then the best engines

of course I imagine that millions of dollars in prize funds must make economic sense for the match organizers
Still...It Just Seems Inequitable To Me

IMHO Regards
Steve
User avatar
Fernando
Admiral of the Fleet
Posts: 3059
Joined: Tue Jul 31, 2007 4:35 pm
Location: Santiago de Chile

Post by Fernando »

The reason that justify this dance of millions in human tournaments is the simple fact that human player gives something computers cannot: theatrics...People go to see these competences to se e the faces of the guys, the smiles, the shame, the arrogance, the commentary of the experts, etc.
For that reason, being this the only motive to go and see, I sincerely expect that in the future the norvergian guys play...

with bikini, tattoo and a pin in the nose regards
Fern
Festina Lente
kgvetter
Member
Posts: 239
Joined: Sat May 12, 2012 5:22 pm

Post by kgvetter »

The comparison between human players and computer-engines makes as much sense as to compare the capabilities of a (human) weightlifter to a forklift.

I did obtain a correspondence chess IM title in a time when my strongest analysis help was a Fidelity Mach 3 and I remember that to find a way to a win in a 5-piece knight endgame kept me occupied for many nights. Nowadays you feed that position to the tablebases and get get the result in microseconds.

Currently I have a position in correspondence which is won by consultion with the Lomonosow 7-piece tablebases but it does not make me really happy since there is no chance that I would have found the winning sequenzy of moves myself (I would have found it but would have overstepped the 50 moves required to avoid a draw.)

I do not understand the negative remarks about Carsens personality since he wins his games usually by mental strenght, being able not to tire out in the 4th hour of a game, which Kasparov described as one of his own weaknesses...

Gerhard
User avatar
fourthirty
Full Member
Posts: 763
Joined: Fri Dec 06, 2013 8:46 pm
Location: San Francisco

Post by fourthirty »

Monsieur Plastique wrote: I think we should now be turning our attention away from trying to break ELO barriers and actually creating significantly weaker machines and engines. I think these would be far more attractive opponents and for my own part, I'd get far more pride out of snatching a lucky draw against a 2200 ELO engine with hitherto unchartered depths of chess knowledge, then a program which is breaking ELO barriers.
I also agree, but believe there is room for both in the computer chess universe. As computing power increases and chess engines evolve, their ELO ratings will undoubtedly increase.

My wish would be for chess engine authors to put more of a focus on playing natural (and exciting) games against us mere mortals (which HIARCS does better than most). It is much more enjoyable to defeat a computer that puts up a good fight, but makes errors similar to the average human player. Engines that miss an opportunity a few plys deep, or misses a mate during an endgame, rather than just blindly giving away material. I want an engine that screws up like I screw up! Perhaps analysis of a large set of games played by 1400 - 1800 ELO players would identify the most common weaknesses, which could then be built into the engine?

Looking towards Golf as an analogy - As much as I enjoy watching the Rory McIlroy tackle Augusta National Golf Club during the Masters tournament, I know if I attempted to play that course in tournament conditions (black tees, narrow fairways, long rough, lightning fast greens) it would be an exercise in futility. I would have much more fun playing from the blue tees at my local municipal course.

I certainly appreciate the the strength of the engines and will continue to follow the TCEC tourneys, albeit without understanding the deep analysis performed by the engines.

Seems like there would be a market for these "average" chess engines - most of the folks that I see at the local golf course are not playing from the black tees!
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

kgvetter wrote:The comparison between human players and computer-engines makes as much sense as to compare the capabilities of a (human) weightlifter to a forklift.
i know all of the obvious analogies
we still have interest in humans racing each other and yet Sports cars go faster
etc..etc
these are all examples of physical capabilities..not mental capabilities

say what you will...
when i can buy a program for $25 to use on my smart phone that will fit in my breast pocket and it will crush the best human player including the World Champion ..then my interest in the human World champion title is diminished..
period..
i think World Champion engine authors should be paid the same if not more then the human champions

humans as the tool maker(engine author) should not be paid crumbs compared to humans as the chess player
especially as the ratings differences between them grows larger and larger

Close The Economic Gap Regards
Steve
User avatar
pr1uk
Member
Posts: 220
Joined: Wed Apr 02, 2014 2:01 pm
Location: Strood, Kent. UK

Post by pr1uk »

Steve B wrote: say what you will...
when i can buy a program for $25 to use on my smart phone that will fit in my breast pocket and it will crush the best human player including the World Champion ..then my interest in the human World champion title is diminished..
period..
I find that so sad


Peter
King Performance Chess Computer M830
Cubeman
Member
Posts: 460
Joined: Tue Jul 31, 2007 10:50 pm

Post by Cubeman »

I have to agree on most of the points made in this thread, computer chess has reached such levels that I can't follow the games from the the top engines anymore. Seems like chess at these levels require some very deep zug-zwang manoeuvres to pull off the wins as can be seen from the randomly shuffling of pieces for what seems pointless moves.
The World Champs for humans has also lost it's appeal to me, not because computers are better but just that there is way too much theory these days that games are decided most times by home analysis aided by engines and 2nd's . When I was young (25 years ago) I was amazed that the players were still in theory at move 10, fast forward 25 years and we see games follow established lines up to move 25 in many cases. I hope one day that Chess 960 will become more popular as that might rekindle my interest in the World champs once again.
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Post by Monsieur Plastique »

fourthirty wrote:It is much more enjoyable to defeat a computer that puts up a good fight, but makes errors similar to the average human player. Engines that miss an opportunity a few plys deep, or misses a mate during an endgame, rather than just blindly giving away material. I want an engine that screws up like I screw up! Perhaps analysis of a large set of games played by 1400 - 1800 ELO players would identify the most common weaknesses, which could then be built into the engine?
Personally I would grow tired of an engine that made blunders like that. A lot of the PC engines can be confured to play at weak ELO levels but rather than play at the selected level they are all over the place - much more so than a human and devoid of the same strengths and weaknesses for the commensurate human ELO rating. It might be fine for blitz games where little personal time is invested but I don't like the idea for the long games that I prefer to play.

Sure, if I am playing against a human in a classical tournament and they blunder giving me an instant win, it is a let down for both of us but at least I have a point to my name and am one step closer to a podium finish and that elusive, highly coveted Pizza with a Diet Coke (Steve has a great knack for selecting desirable prizes). But if I am at home playing an engine, I might be investing an entire afternoon on a weekend to the game at the expense of other things. For the machine to make a bad blunder - and by that I mean that I am specifically expecting one to occur each and every time I play it - really ruins the afternoon for me. For this reason I like to distinguish outright blunders from sub-optimal moves. Sub-optimal moves are exciting because you need to come up with a sequence of moves to take full advantage of it. That is exciting. Walking into a three ply fork or skewer is not exciting, nor is a machine that sends a pawn into suicide territory in an otherwise dead-even endgame simply because it lacks any knowledge and cannot think far enough ahead.

Anyway, for anyone wanting one of the most humanlike entities out there for the rating, pretty much nothing goes past the Novag Star Opal. You don't need an engine. Just buy the $100 machine instead if you are a hobby 1500 to 1600 ELO player. Quite simply it is the most humanlike for the rating machine I have ever owned and played against. If you set it to the lower average levels it will often make bad blunders. As the levels go up the blunders and inaccuracies decrease. Set it to the higher fixed time levels and it behaves itself with a blunder rate commensurate with a human of that lower level club strength trying their best in a tournament game. This is exciting chess - it's like the scratchy tickets. You never know if you have pulled a $20 win and with the Star Opal you never know if and when you are going to get a mistake from the machine for you to exploit or not.
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
fourthirty
Full Member
Posts: 763
Joined: Fri Dec 06, 2013 8:46 pm
Location: San Francisco

Post by fourthirty »

Monsieur Plastique wrote:Personally I would grow tired of an engine that made blunders like that.
Monsieur Plastique - I believe we are both promoting a similar concept.

I don't find much joy in playing against a machine (or engine) that makes obvious "blunders", such as blindly giving away material, when it is playing at a Handicapped level. I view a blunder as a move that is an instantly recognizable error (by me).

I am also a proponent of a engine that would make sub-optimal moves. I would consider the examples that I listed above (missing an opportunity a few plys deep or missing a mate during an endgame) as making a sub-optimal move. Trying to finish a game against an engine that has access to a one terabyte 6-man endgame tablebase is not my idea of fun! Although I do love having access to this feature in HCE for analysis...

When playing against another human, there are certainly times when my opponent or I does make an obvious blunder. However, more often, the majority of the wins/losses are a result of a string of sub-optimal moves - which I normally don't even recognize while they are happening!

Thanks for the tip on the Novag Star Opal. I don't own one of these units but will definitely keep my eye out for one.
Post Reply