The Sorcerer's Apprentice

[Lyudmil Tsvetkov]

Nothing can stop me.
Only thing that can stop me is when Louis says, Let's break it.

I don't want to stop you neither Louis, but could the two of you once in a while consider to play these "games" by PM if you're simply not able to come to a presentable conclusion of any home- analysis on your own?

And if you then both think they are of common interest, show only the parts with any kind of outcome without all that slapdash trial and error and trial again ?

I can understand that with each and every new public belly flopper your will to show you're the great Zauberlehrling grows (the analogy in Goethes sense ist really good by the way, in the sense of "In die Ecke, Besen, Besen, seid's gewesen. Denn als Geister ruft euch nur zu seinem Zwecke
erst hervor der alte Meister." ) but that's a vicious circle too, Luydmil.

Sincerely your Baumwolle

I even do not read your posts anymore.

Do not be funny.

There are basically 2 types of test positions: normal test positions, and mine with Louis's analysis.

In normal test positions you have:

- several people providing engine analysis to certain depth, probably not very big
- usually engine output is accepted as it is, whether people understand it or not, only sometimes there is a discussion
- usually standard or very standard positions that you see by tens of thousands daily

In mine with Louis's test positions you have:

- engine analysis provided by the best engine currently in the world, with much longer thinking time, and then this engine analysis is further checked and rechecked with subsequent tries
- attempt to go to the bottom of it, and possibly even discuss some eval terms that might be of use to engine authors; actually SF and some other engines have benefited somewhat from those threads and the eval ideas discussed in them, so, if you do not like the threads, why are you using latest SF?
- cherry-picked very uncommon positions that most of the time engines do not understand; that is the whole point of it, if you have not understood it bu now - that engines and their authors might have a glimpse of what is possibly wrong/still not perfect with their creations

I would say that a position like the one above is a real gem, extremely rare to find, but you should have a taste for it. Ne sus proices margaritas!

So I really do not understand you. I can do the following thing:

- post positions with an easy win engines will easily recognise; whom will that interest, there are many such positions
- post positions that engine absolutely will not understand, yes, I can easily do that, all top engines will be completely clueless and I will be able to win in a couple of moves, but, again, whom will interest that, the solution is given and that is the end of it
- choose positions that are unbalanced, with mutual chances, like the one above, when you have hard time to decide who has the advantage and each side should do its utmost to prove its point; that is what I am actually doing and I think it is the best way to proceed, as the puzzle is still not solved and puzzles are interesting

Concerning PMs, did I read something about PMs?, yes, definitely, we will do a PM (post mortem) after the game.

And please do not forget one other thing: SF has improved by 70 elo on Louis's 16-core since we last sparred in fall of last year - 30 elo from software on single core, and another 40 elo from SMP scaling from 8 to 16 core. If you think it is an easy task to handle such an engine, I assure you it is not precisely like that.

[Lyudmil Tsvetkov]

I think I'll finish the game in progress, and then take a break from this practice. Or follow your suggestion to play via PM.

Please don't break. I'll miss those games or whatever. Talkchess will not be the same without them.

Thanks for the moral support, Lyudmil.

[Lyudmil Tsvetkov]

I don't want to stop you neither Louis, but could the two of you once in a while consider to play these "games" by PM

I've often worried that these "games" are an inappropriate use of the forum. For me they are easy to "play". I just relay moves to and from Stockfish, while I work at other activities. I must admit that I don't have a lot of time to really think about the positions that arise. What has kept me from stopping is my belief---perhaps very wrong---that some others find the games entertaining. And also my knowing that no one is forced to look at the thread.

Hi, Louis.

I have to admit that I follow these threads religiously, rarely for the chess itself anymore, but increasingly for the schadenfreude as Lyudmil's overweening claims are repeatedly dashed against the rocks of Stockfish.

That you respond with move after move and accept do-over after do-over, so steadfastly and meekly, provides the means for this entertainment and is very admirable. My only fear is that your humility will somehow rub off on Lyudmil, and the show would be over very quickly thereafter.

What a language.

What are you: Russian, Russian expatriate living in America?

You said it quite right: you feel Schadenfreude, a German term with extremely negative connotations - people feeling Schadenfreude have something destructive about them, do not they? They do not want to appreciate the real value of things, but are rather happy when other people's hopes, aims and expectations prove to be a failure.

Why do not you do something creative and positive yourself, instead of simply criticising?
Then we will see what you are actually able to do.

Again, I can post positions engines completely do not understand, and win every single game, but whom will that interest?
That is why I am posting games that are unbalanced and yet undecided.

Who is going to watch Barcelona-Real Madrid, if they know the score in advance?

I wonder what you really expect from me: to win easily every single game against SF on 16 cores and more than 30 minutes thinking time per move?

If you can do it, why do not you play a game against SF yourself?

[Lyudmil Tsvetkov]

Nf1

[d]r3nrk1/ppq1bppb/1n5p/2pP1P2/2P1P3/1PQ2N2/P5PP/R1B1RNK1 b - - 0 8

SF plays f6 with evaluation of 53 cp for Black.
[d]r3nrk1/ppq1b1pb/1n3p1p/2pP1P2/2P1P3/1PQ2N2/P5PP/R1B1RNK1 w - - 0 22

Your move when you return.

Ng3

[d]r3nrk1/ppq1b1pb/1n3p1p/2pP1P2/2P1P3/1PQ2NN1/P5PP/R1B1R1K1 b - - 0 9

Louis, please do not listen to them, you have been doing great job all along, if someone is not able to appreciate, that is only their fault.
The position we are currently analysing is one of the most interesting positions possible; what more could you actually wish than tackle at long such a position?

I mus admit that I do not have an easy task either: SF has improved tremendously from the time we sparred in last November, some 70 elo overall with the SMP scaling from 8 to 16 cores included, so it should make less mistakes.

I think it played very accurately the game until now, picking all the best moves, so I really do not have an easy time. If it had gone wrong only once, the game would have finished, but it did not.

People around get confused by the 0.0 scores SF showed initially turning to black advantage; well, first, those are bogus scores, and second, I avoided the draw in other lines with Qc3 on purpose, white is playing for a win after all, not so?

I am convinced white has the advantage, but how can you explain this to people who only trust engine scores?

I think there is nothing more interesting than an unbalanced, strenuous game: it takes the best out of both players, and the end result in terms of quality of play should be, of course, optimal. The more strenuous a game is, the higher the quality.

[Robert Flesher]

Nothing can stop me.
Only thing that can stop me is when Louis says, Let's break it.

I don't want to stop you neither Louis, but could the two of you once in a while consider to play these "games" by PM if you're simply not able to come to a presentable conclusion of any home- analysis on your own?

And if you then both think they are of common interest, show only the parts with any kind of outcome without all that slapdash trial and error and trial again ?

I can understand that with each and every new public belly flopper your will to show you're the great Zauberlehrling grows (the analogy in Goethes sense ist really good by the way, in the sense of "In die Ecke, Besen, Besen, seid's gewesen. Denn als Geister ruft euch nur zu seinem Zwecke
erst hervor der alte Meister." ) but that's a vicious circle too, Luydmil.

Sincerely your Baumwolle

I even do not read your posts anymore.

Do not be funny.

There are basically 2 types of test positions: normal test positions, and mine with Louis's analysis.

In normal test positions you have:

- several people providing engine analysis to certain depth, probably not very big
- usually engine output is accepted as it is, whether people understand it or not, only sometimes there is a discussion
- usually standard or very standard positions that you see by tens of thousands daily

In mine with Louis's test positions you have:

- engine analysis provided by the best engine currently in the world, with much longer thinking time, and then this engine analysis is further checked and rechecked with subsequent tries
- attempt to go to the bottom of it, and possibly even discuss some eval terms that might be of use to engine authors; actually SF and some other engines have benefited somewhat from those threads and the eval ideas discussed in them, so, if you do not like the threads, why are you using latest SF?
- cherry-picked very uncommon positions that most of the time engines do not understand; that is the whole point of it, if you have not understood it bu now - that engines and their authors might have a glimpse of what is possibly wrong/still not perfect with their creations

I would say that a position like the one above is a real gem, extremely rare to find, but you should have a taste for it. Ne sus proices margaritas!

So I really do not understand you. I can do the following thing:

- post positions with an easy win engines will easily recognise; whom will that interest, there are many such positions
- post positions that engine absolutely will not understand, yes, I can easily do that, all top engines will be completely clueless and I will be able to win in a couple of moves, but, again, whom will interest that, the solution is given and that is the end of it
- choose positions that are unbalanced, with mutual chances, like the one above, when you have hard time to decide who has the advantage and each side should do its utmost to prove its point; that is what I am actually doing and I think it is the best way to proceed, as the puzzle is still not solved and puzzles are interesting

Concerning PMs, did I read something about PMs?, yes, definitely, we will do a PM (post mortem) after the game.

And please do not forget one other thing: SF has improved by 70 elo on Louis's 16-core since we last sparred in fall of last year - 30 elo from software on single core, and another 40 elo from SMP scaling from 8 to 16 core. If you think it is an easy task to handle such an engine, I assure you it is not precisely like that.

This is the post in which you show you true colors. You make wild claims and when Peter (or others) ask for proof, (winning lines, pgn) for which you cannot/never provide, you state you no longer read his posts. It provides a means to make you look stupid and arrogant. The "troll" branding that at least one highly respected member has already labelled you seems to fit the bill! Enough of the bullshit, if you say it's winning, be prepare to show proof. Capiche?

[peter]

I even do not read your posts anymore.

Well, for not reading them your answer is quite long, thank you for that, I yet would read what I'm answering to first, a thing that I can recommend as for playing chess too, look at the possible moves of the other side too before answering.

There are basically 2 types of test positions: normal test positions, and mine with Louis's analysis.

That sound's a little symptomatic as for your kind of attidude too, Lyudmil, you really seem to think, you have invented the wheel of test positions and of testing them for new, I don't think your kind of "testing" positions of your choice (obviously especially as for your chances to win them against the engine with the same engine's help) new, it's simply bad testing-craft you're using, you don't test lines and outputs as for the relations of the evals, you simply try and fail move by move hoping for a failure of the engine.
At the end you don't even reevaluate the moves you and your "opponent" made.

What you're doing is simply a bad in-between of engine- testing by counting points at the end of games and position- testing, which is something very different.
As for simple counting full and half points, the amounts of games are much to much to little, even for single positions only and some kind of theme-tournament, because you'd have to judge statistically in the end.

Concerning PMs, did I read something about PMs?, yes, definitely, we will do a PM (post mortem) after the game.

Of course I meant private messages (of any kind) with PM, but a post mortem analysis of your games with SF was something I yet would recommend to you too as a method of keeping the postings shorter (you could first reanalyse and post then) and to learn from your own blunders too.
You always jubilate finding the engine to blunder, I never ever read a single word from you, when you once again goofed a "game" by yourself, maybe we could have a deeper look at these mistakes, misevalutions, clear blunders, little headaches and fatigues of yours now and then too, at least afterwards?


And the way to spare this to you and to us was simply to look deeper into the positions and its lines before posting.

Louis would help you probably with your home- analysis in other ways of communication (what about some kind of PMs? ) too and you'd see what happened against a good hardware- time with your plans and moves, maybe you'd even come to certain usable and testable variants before you start bragging again "this position is clearly won" the next time.


One thing I learned about engine- testing, and believe me, I'm doing it since quite a while, when it comes to evals, never ever judge single ones of single positions and single moves without comparing them to the lines of the output of the engine and alternative ones you have judged for yourself at first.

What I'm simply missing from your human "output" all the time, is any kind of testing- line together with your now and then very imprecise and now and then provable wrong evalutation of the positions you bring along, as interesting as the positions now and then are. No wonder, there is no relevant evaluation of a position without at least a single pv according.

And the positions themselves: now and then they are simply less interesting then you think and know by yourself (interest is something quite subjective always of course, but there isn't any without being with the thing of interest for a while, "inter"-"esse"=being "inter") and now and then they are even more interesting than you know but in quite a different way.

Often enough, they are dead remis from a certain point and you don't have a clue about that till you rode your dead horse to death several times and several races more, that gets on our nerves mostly, I really admire Louis' patience in these neverending "games" very much.

In the cases, when positions turn out being of some other eval than you thought at first glance, you sometimes could be less stubborn and more able to admit that, otherwise you once and again simply look like a sore loser, Lyudmil, with all due respect for your great wins. (with the help of SF, pity I don't know others probably)

[lantonov]

http://german.about.com/library/blgzauberl.htm

is one of the trials I found to translate the whole poem, I yet don't like such, you cannot really translate poetry

This makes much more sense, thanks.

[Lyudmil Tsvetkov]

Capiche?

Capisci/capisce?

Mi dispiace, non capisco.

[peter]

Capiche?

Capisci/capisce?

Mi dispiace, non capisco.

Take some care, Lyudmil, even more than pumpkin chapiche has as metaphoric meaning in English.


http://en.wiktionary.org/wiki/capiche

At this one link

http://www.urbandictionary.com/define.php?term=Capiche

you have to scroll down a little, to see, what I mean.

And if you then still haven't got the full dangerous meaning of capiche, watch some Francis Ford Coppola- films, then you'll know better.

[lantonov]

What I'm simply missing from your human "output" all the time, is any kind of testing- line together with your now and then very imprecise and now and then provable wrong evalutation of the positions you bring along, as interesting as the positions now and then are. No wonder, there is no relevant evaluation of a position without at least a single pv according.

I look at this from another angle: as human thinking vs. machine thinking. It's not just any human -- Lyudmil has abundantly proven his deep understanding of chess theory, building on all accumulated human chess knowledge over the past centuries. It's not any machine either -- SF is the best chess program in the world (or one of the best, not to anger Komodo fans). Lyudmil's share in SF's development is significant. Just look in Fishtest and you will see dozens of patches inspired by his ideas here in Talkchess, many of which are successful, contributing real Elo in SF strength coming from improved evaluation.

Differences in thinking and approach are also interesting. Humans tend to express their ideas in imprecise language while engine gives PV and numerical scores. Nevertheless, Lyudmil has the knack to find the weak spots in the engine in positions and elements which are difficult to program (e.g. closed positions). His most important ideas are not implemented precisely because they are foreign or unusual to the machine way of thinking.