Everyone on mersenneforum vs. Stockfish

[Brian-E]

Originally Posted by axn

I think "less than 24 hours" refers to time allotted to the machine to make the move, not for the humans?

Ahh! In that case, no complaints from me. Thanks.

EDIT: You're quite right. I even quoted MooMoo2's post where it says exactly that. Somehow I seem to have lost the ability to read.

Originally Posted by gd_barnes

Likely the only chance that we would have would be to beat it will be with long-term quiet strategical positions.

Agreed.
Which is why I stated, "...At some point I would like to overwhelm Stockfish with a complicated board."


I'll explain myself.
I will assess the opposition (Stockfish), assume nothing and exploit/create weakness (as opposed to a trap/meaning a trap) as I recognize it.
This means that I will be aware, as much as I am able, of the mathematics and algorithms (reasoning/adaptive processes)
in whatever form they exist and manifest. I can almost guarantee you are a better chess player than I am. I will only
state that I play fair and I play to win but am a gracious loser. But if you're a prick, gloves off and brass knuckles on.

The following means nothing but like "Galaxy Quest's" historical documents here is some of my chess background:
In Gr.6, learned to play chess by pushing labelled paper squares on a cardboard slab; at 12 in High School, in the
only tournament I ever played, I lost to everyone except the regional champion. I used a Queen sacrifice to beat him; as a junior in
University in '79 I wrote a [bad] chess playing program in Fortan 77 using keypunch cards; I've read Shannon's papers; a recent
issue of Chess Magazine featured an article on Chess Cheaters which is interesting; I like Fritz 14 (fully loaded),
play when I can and/or show the rudiments of the game to those who haven't played but are interested. GO is an excellent game.
Here is a good link which I came across many years ago (when they were first starting) to improve my poker game(s): https://webdocs.cs.ualberta.ca/~games/

\\
Joshua: Greetings, Professor Falken.
Stephen Falken: Hello, Joshua.
Joshua: A strange game. The only winning move is not to play. How about a nice game of chess?
\\

Now you tell me, what DON'T you know about playing an unbeatable game of chess?

I apologize for the last sentence, uncalled for.
My point there was that you cannot know what you don't know, you can only infer.

[gd_barnes]

Originally Posted by jwaltos

Agreed.
Which is why I stated, "...At some point I would like to overwhelm Stockfish with a complicated board."

I am not trying to be mean. I'm just stating facts. Here again...your first sentence makes no sense. I said the only chance that we would have to beat Stockfish was with quiet positions You initially agreed and then contradicted yourself (again) by stating that you'd like to overwhelm it with complicated boards. Not to be rude...Is English your native language? In English complex boards implies tactical positions not quiet strategical positions.

Regardless of your math/programming background, LaurV also thought that your statement was a joke. I'm not the only one. You kind of lost credibility with the "try the 2-move checkmate" statement. You cannot overwhelm a strong program with a "complicated board" which implies tactical complex positions. See http://mersenneforum.org/showpost.ph...0&postcount=20.

[Brian-E]

Well....

"Complex" is a complex word, as is "complicated". It will not mean the same thing to a human playing chess as will to a human programming a machine to do that.

The minimax algorithm (add appropriate cut-off mechanisms including killer heuristics), which is perfect for playing tactical positions where brute force calculation is required, is nearly useless in a strategic positions where general considerations are needed. So to a chess playing algorithm you could indeed call a strategic, positional type of game "complex" or "complicated".

[Batalov]

I've run a chess robot on a certain Russian mulitplayer site for quite a few years ( http://www.playelephant.com ; in fact I can recommend it to players with any language - they now have a decent internationalization for the interface; as for the chat with users - you don't have to use it). Hacked its clients' guts and then let the site owners know, after a week of testing. We had a good laugh and got very well acquainted. Since then they immediately ramped up the client's code security and of course integrity check upon connection but by our agreement let the only one hacked client connect - mine, and so I ran the robots for a few years until I got bored (ca. 2006-2011).

Because the game client was fully functional but had made all moves automatically (it was a proxifying adapter to a child process, one per each game type), I had frequently sat in, observed and chatted from within the robot's chat window. There were significant kibitzer crowds at some high level games. IMs present at the club occasionally played for show (and of course with rating changes disabled for these matches :-)

So, I've seen people draw against my chess robot and even win (but invariably only when they insisted on {1-2minutes}+0s non-Fisher, quickly built a very closed position and then would tediously win on time because they were in Moscow with a 10ms r/t to the server which is in Moscow and the robot was in San Diego with a 450ms r/t). In a closed position, the robot was very timid and frequently just shuffled a rook aimlessly for all of the middle game (and obviously there was no endgame). If the human opponent was not observant and didn't dance around well enough, the server would register a 50-move rule draw or a triple repetition draw. But I did see people win. About 10-12 times out of 28,000+ of games... See
https://www.playelephant.com/user/stats?user=Crafty ,
https://www.playelephant.com/user/stats?user=Happy,
https://www.playelephant.com/user/stats?user=Grumpy
(I had other gnomes as well, Sneezy, Bashful etc...)

And the engine was the relatively weak Crafty (I always only used open-source robot engines*), not Stockfish, and even then no one ever drew with a 1s Fischer correction. So I am very skeptical. The more time you give to the engine, the worse it is for humans.

____________
*Here are to the best of my recollection the engines I've used
chess: GnuChess, then Crafty
antichess ("poddavki", Rule 17), CrazyHouse: Sjeng-opensource-11.2
reversi/othello : edax
gomoku : Tito (the only closed-source binary, but with well-defined API)
backgammon/nackgammon : GnuBG
Go : MoGo, GnuGo
Highest rating places for all game types were #1, except Go where it was at best #4. Right now you will see no #1 rating place for chess and other games because the club implemented activity cutoffs (one has to play, perhaps once a week, to be shown in rating tables, and obviously the robots do not pass this; they haven't played for 4-5 years)

[Batalov]

P.S. You will find that these days, in chess hall on playelephant you will be able to count player using one or two hands. It's an unfortunate side-effect of the commercialization of that site. It is a boring entertainment site these days, but it used to be the site and the club. It had a gentleman's club atmosphere, it was free and democratic and had great attention to the quality of both practice and tournament games, which had human referees and the players of highest level. In contrast now, it has slanted towards appealing to the masses and very weak (but paying) "customers" are their visitors these days. The attendance is up to 1,000s and even 10,000s of users at some times of day (and it used to be hundreds-thousands) but the quality is all gone. The automatic tourneys that start every 5 minutes and are refereed by the server automatically are no fun, really... and there are almost no conversations except for occasional blaming rants from opponents (that's of course in 3-4-player games; invariably, from people who think that they know the game better than others, the usual Dunning Kruger in action). All of my old friends are gone from that server and that's no longer a place that I have to go to chat with my sister when she was playing Durak or Dominoes while waiting for another tomography patient*. I have Skype and WhatsApp now for that.

Needless to say, you won't find IMs or GMs there anymore. Adjust your expectations. You can still play. I play preferans once a month -- just to remember how.

____
*yeah, there, now you know what the X-ray specialists of highest qualification occasionally do on their PC connected to a Siemens 4-ton tomograph. The 24-hour shift is brutal if you have no fun. Hopefully she is still not playing a game when she presses a button and a metallic voice behind the led-enforced glass says 'Hold your breath'.. .'Breathe now'.
No, I am only kidding. I've actually had a privilege to see her working, for hours; and even a couple of nights (we were waiting for something bigger that was happening in parallel at the same hospital where she works; we had days and nights to kill). 3D reconstructions of someone's spines, abdomens, skulls were built in front of my eyes with a few master strokes, virtual data slices revealed fractures, masses, reconstructed vertebrate, cracked screws in titanium scaffolds, hydrocephaly (take your pick); write-up reports were generated in minutes... 'Roll in the next one!'

[Brian-E]

Serge, I would say that your machines' results are a tribute to the effectiveness of the cheating detection on playelephant.com. Presumably the programs would have dropped half- and full points here and there to opponents using their own chess playing software. On both of the live chessplaying sites on which I have spent significant time, players are caught cheating very frequently (and have their accounts immediately removed in one case, or converted to a computer account in the other). But I find it hard to believe that everyone doing it gets caught, especially if they only do it occasionally.

[Batalov]

It is very true (and sad) that there are many hidden "robots", that is people who pretend to be strong for whatever psychological reasons they may have. I have forgotten to address that in the above reminiscence, but now that I thought of that, indeed, initial reaction to robots was curiosity from most users but a very strong resentment from a few. And we can see why! That was a factor, indeed, -- even though, as we have pre-discussed with the site's owners before I even launched the robots that they will have a clear label - "This is a robot" (not only it was added to their userpages, but the server programmers specifically created a user category and the label "Crafty [Robot]" was added to every user interface label (it showed straight as a continuation of the username at every vacant table where people would inadvertently sit at and could immediately realize if that fit their interest or not - to play against a robot, or next to the opponents name if the game was in progress. At all times.). In fact this category was added distinct from an already existing category that dishonest and calculated-out human players would get; it was "[Proga]" (i.e. Russian shorthand for 'a program' or 'a progger', rather) and sounded (and was designed to sound) deprecating. Most "Proga" users, once revealed by analyses, were labelled, shamed and they usually would leave - save for a few weirdos who wore that label with pride. Very few people played with them. Their tournament results were nullified.

An interesting situation would always arise when some of those pride-sore "Progas" would come in to show what they are really worth (as if!) to play against robots. That was always very funny. Because "Progas" had to have split screen (or two PCs) and humanly limited reaction on transferring moves - they always lost. The most pathetic ones were dronely paced at, say, 4-5 seconds per move (including completely obvious ones! this is partly how they were invariably discovered). The most talented ones used hybrid strategies and skills and would switch to playing solo without a program for forced moves and endgames, noticeably faster. Maybe some of them were in fact, human pairs.

Only one of them once or twice won - he would come during my San Diego nights when robot was unattended (and was left at a spectator's request at, say, 4min + 1s option setting) and would bombard a robot with game after game with absolutely repeated games using a stronger engine and deliberately dropping all games of one color and any unwanted debuts. His summary score was disastrous, perhaps a 1000-:3=:1+, but that's because his goal was to win once. Which is not all that undoable if one would unleash Rybka against Crafty and has enough time to transfer moves. So, there, that was another questionable 'win at all costs' strategy in addition to winning on time in a dance-around position (Crafty engine didn't have an escape strategy from longplay; maybe it has now, I don't know). I am quite sure that now, after another 5 years, new engines have meta-strategies to identify and deal with a situation where its side needs to take initiative to drift away from a boring repetition rot.

I speak Khoisan and Linear B and am contrite over my English fluency shortcomings. Having lost all credibility where none existed before via a democratic process moved me to tears. Semantically, I am confused regarding your statement concerning "quiet strategical (?) positions."
Is this a sexual position of some sort? I have burped, belched and farted (but no queefing) LOUDLY during many a strategic session. Just funnin' with you GB, however, you would not make a good John Connor - not cool.

My communication skills (or lack thereof) have gotten me into trouble before so I will work on that hopefully without becoming pedantic.

During my online poker playing years ago bots existed. Rather than cry about it I studied them and built my own (did not let them loose though). The little chat boxes were wonderful for putting people `on tilt.` If StockFish were sentient perhaps a little chatbox would make games more interesting..just sayin'.

Here is an interesting link regarding minesweeper: http://web.mat.bham.ac.uk/R.W.Kaye/minesw/minesw.htm - there are analogies with chess engines. Anyways, I'm here to learn and play and not waste anyone's time with bullshit regardless of how sweet it smells.

[MooMoo2]

The poll just closed. The most popular move was 1.e4, and stockfish responded with 1... e6. The FEN is:
rnbqkbnr/pppp1ppp/4p3/8/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - 0 2

A poll for move 2 will be posted in a few days.