How can you tell if you're playing against a computer?

Originally posted by wormwood
same for 'memorizing computer lines' which gets killed by combinatorial explosion.

Prior to playing any games, suppose I decide that I'm going to start playing the Sicilian Najdorf and I analyse many GM games with an engine. I save my analysis to file. Somebody else expects to play against the Najdorf as White, and they conduct likewise upfront analysis. We then play each other on RHP. Are you saying that we're not allowed to reference the "pre-existing" analysis? Are you saying that it's unlikely our computer analysis will be the same?

Originally posted by Varenka
Prior to playing any games, suppose I decide that I'm going to start playing the Sicilian Najdorf and I analyse many GM games with an engine. I save my analysis to file. Somebody else expects to play against the Najdorf as White, and they conduct likewise upfront analysis. We then play each other on RHP. Are you saying that we're not allowed to reference sting" analysis? Are you saying that it's unlikely our computer analysis will be the same?

This is what it says in the TOS: While a game is in progress you may not refer to chess engines, chess computers or be assisted by a third party. Endgame tablebases may not be consulted during play but you may reference books, databases consisting of previously played games between human players, and other pre-existing research materials.

I'd think your analysis in a file would count as a database and I'd guess the "between human players" part might now be debatable. This probably renders most databases illegal on this site, especially if said DB includes modern CC games. The same would likely go for most books on opening theory. GP will be pleased!

As to whether or not your opening analysis will be the same, I suspect it will depend on what engine you use, how long you let it run, what kind of iterative process you use to progress the analysis and so on. As an example, Houdini likes the following opening sequence: 1. e4 e5 2. Nf3 Nf6 while Stockfish likes 1. e4 c5 if my five minute run both in tandem quick test is anything to go by.

Originally posted by Diophantus
I'd think your analysis in a file would count as a database and I'd guess the "between human players" part might now be debatable. This probably renders most databases illegal on this site, especially if said DB includes modern CC games. The same would likely go for most books on opening theory. GP will be pleased!

Firstly, I'm sure we all agree that books with opening theory are perfectly fine on RHP - otherwise the number of "cheats" on here has just risen dramatically! 🙂

Secondly, if I print off my file of analysis prior to starting a game, in which way is it any different from the other opening books I have on my shelf? I have books where the author used an engine to check analysis and these are legal "pre-existing research" on RHP.

Originally posted by Varenka
Firstly, I'm sure we all agree that books with opening theory are perfectly fine on RHP - otherwise the number of "cheats" on here has just risen dramatically! 🙂

Secondly, if I print off my file of analysis prior to starting a game, in which way is it any different from the other opening books I have on my shelf? I have books where the author used an engine to check analysis and these are legal "pre-existing research" on RHP.

First, I would be willing to bet that we don't all agree that books with opening theory are perfectly fine. I have been asked if I am cheating by looking at a book so there are a few who don't think books are at all fine. I do think that tiny minority are unlikely to get very far with the notion that book use is cheating but that doesn't stop them holding that opinion.

Second, there are a few problems with allowing computer analysis in any form under the current set of rules. Is a book containing computer analysis legal pre-existing research? If the book is then so is the database and yet the TOS says no. How do we define "pre-existing"? Must some particular time period pass before you can use your computer analysis? If the only requirement is to have it printed out then one can simply perform computer analysis and then print it out before making a move. That doesn't seem right somehow.

To be honest, I think it would be impossible to ensure that any database of reasonable size does not contain any non-human input. Even the RHP DB must contain some games with a computer element simply because people who are represented in that DB have been banned for engine use. Considering how often the RHP DB gets updated it is unlikely that admin goes through it removing all games played by banned players.

Originally posted by Diophantus

I would be willing to bet that we don't all agree that books with opening theory are perfectly fine

I wasn't comparing opinions; I was discussing what is legal or not on RHP. The TOS says "you may reference books" and there is no mention of opening books being excluded.

Is a book containing computer analysis legal pre-existing research?

Yes, because the TOS says ""you may reference books" without excluding the above.

If the book is then so is the database and yet the TOS says no

I don't follow your logic. The TOS says "databases consisting of previously played games between human players" so that applies to databases, but I see no corresponding statement for books.

How do we define "pre-existing"?

The analysis existed before the game started.

If the only requirement is to have it printed out then one can simply perform computer analysis and then print it out before making a move.

No. It's not pre-existing if it's motivated by a current game. In your statement, the analysis is being done when the position exists - not pre-exists.

And we have more than just these scenarios where the TOS allow "room for maneuver"

How do you cover someone who plays on several sites, some that allow consultation with engines and some that don't. Does analysis produced for one game on an engine allowed site become available for the same player on a non engine site. What if the first game is not available for public viewing, as ICCF for example, until the game is finished, even if the critical position is reached before the RHP game. The excuses that can be used to cover engine use go on and on.

Should you lodge a copy of your printed analysis with a commissioner for oaths before a game starts?

I understand that "serious" correspondence players will research their opening lines quite deeply, even at amateur level, and far more so than equivalent OTB players do, and that the origins of available material is now quite blurred. I also understand that an engine is probably the only means an average player may have to "quality assure" master level games in their database that they may wish to follow in a correspondence game.

I fully empathize with players who say that such a person is not using their own brain and chess skills, and therefore questionable ethically.

Originally posted by Varenka
Is a book containing computer analysis legal pre-existing research?

Yes, because the TOS says ""you may reference books" without excluding the above.

If the book is then so is the database and yet the TOS says no

I don't follow your logic. The TOS says "databases consisting of previously played games between human players" so that applies to databases, but I see no corresponding statement for books.

Originally posted by Varenka
Secondly, if I print off my file of analysis prior to starting a game, in which way is it any different from the other opening books I have on my shelf? I have books where the author used an engine to check analysis and these are legal "pre-existing research" on RHP.

This is the bit I was answering here. You have tried to make the rule apply to a file, in effect a database, that you have printed. If printing makes the resource a book I can reference anything I print out. There isn't a corresponding statement for books but if I can just print (or claim to have printed) and then count it as a book, I can get around the rule limiting databases to human only games. Which is what you were asking about in the first place. So how is that different to any other book? Depends on whether or not you accept that your printout is equivalent to a book. And whether or not others accept the equivalence I suppose.

So, what do you reckon? Can you just print your computer analysis and claim it is a book or should you be tarred and feathered for dubious behaviour? More important, how would we know you were up to no good?

Originally posted by Diophantus
If printing makes the resource a book I can reference anything I print out.

I didn't mean to imply that anything that comes out of my printer is a "book". I meant, how does a print-out of computer analysis (done before the game) differ from the opening books I have on my shelf in terms of the information content? I didn't mean they become the same in all ways. e.g. I usually regard a book as being something that is published, etc. whether that is a formal definition or not.

The key point is: what is the limit, if any, of "pre-existing research materials"? Where in the TOS does it limit this? And if I have a single game that is heavily analysed and saved to file, then I don't regard that as a "database".

Originally posted by Varenka
I didn't mean to imply that anything that comes out of my printer is a "book". I meant, how does a print-out of computer analysis (done before the game) differ from the opening books I have on my shelf in terms of the information content? I didn't mean they become the same in all ways. e.g. I usually regard a book as being something that is published, etc ...[text shortened]... e that is heavily analysed and saved to file, then I don't regard that as a "database".

and I'm not arguing with you. There is no limit in TOS other than our own interpretation of what that bit of the TOS says.

I think what is needed is how those in charge interpret the rules. It is all well and good you saying that you do not a single heavily analysed game as a database but what if the intention is to ban the use of that resource if computers were used? We need guidance on what was actually intended as the current situation almost allows us any interpretation we like.

Originally posted by Varenka
Are you saying that it's unlikely our computer analysis will be the same?

well, do engine vs engine matches produce identical variations and predictable exact end results for individual games? no. they vary greatly, even if you pit the same exact engine against a copy of itself on identical hardware.

they probably won't stay the same for many moves. you'd need to use same exact hardware, software (including EVERY process running since boot), configuration (again, for everything, including BIOS), AND time it all down to one cycle accuracy to stay in the same line and same DEPTH in every searched branch. and still a random bit would flip on one machine but not the other, identical hardware components would have individual variations off the assembly line etc etc... the lines would soon diverge.

not to mention how unlikely it is to play against another opening fritzer who'd guide his engine through the same line and not some equal sideline he prefers for subjective reasons.

Originally posted by Diophantus
as the current situation almost allows us any interpretation we like.

no. the TOS is very clear.

books are allowed.
human game databases are allowed.
pre-existing engine analysis is allowed, with the explicit exclusion of engine game databases.

tablebases are NOT allowed, and neither is any kind of 3rd party assistance on an ongoing game.

end of story.

Originally posted by wormwood
well, do engine vs engine matches produce identical variations and predictable exact end results for individual games? no. they vary greatly, even if you pit the same exact engine against a copy of itself on identical hardware.

The "top 3 matchup" approach relies upon someone being able to run an engine as part of post-game analysis and see that there is a high level of matchups. And this is in cases where we don't know the initial player's hardware or software. If it varied too greatly - even in the case of blatant cheats - this method could never work.

So how can the "top 3 matchup" sometimes result in a high level of matchups, but yet two individuals analysing the same opening are not prone to producing similar analysis?

Originally posted by Varenka
So how can the "top 3 matchup" sometimes result in a high level of matchups,

short game, forced lines, can occasionally end up even in a 100% matchup for a single game. it's very rare, but not unheard of. which is why single games mean nothing.

you should really know basic stuff like this by now.

Originally posted by wormwood
short game, forced lines, can occasionally end up even in a 100% matchup for a single game. it's very rare, but not unheard of. which is why single games mean nothing.

you should really know basic stuff like this by now.

You're avoiding the question.

You try to claim that engine analysis varies greatly, but yet it is fundamental that the same analysis can be reproduced as part of the "top 3 matchup" method. i.e. if a cheat uses an engine to suggest Nf3, then I need to get my engine to do likewise in order to gather evidence (sure, not on every move, but on a high enough amount of moves).

But yet, if I use an engine to suggest Nf3 as part of opening analysis prior to starting a game, you're saying that engine analysis varies too greatly for another person to also produce the same analysis. This contradicts the above.

Originally posted by wormwood
no. the TOS is very clear.

books are allowed.
human game databases are allowed.
pre-existing engine analysis is allowed, with the explicit exclusion of engine game databases.

tablebases are NOT allowed, and neither is any kind of 3rd party assistance on an ongoing game.

end of story.

Would you view a database that contains no more than 1% engine games (i.e. few games but still there nonetheless) as an engine database? What if some of those games are human v computer? How about a database of CC games played since 2000? The latter will certainly include a great deal of engine input and yet it is nominally entirely made up of human v human games. That's what I mean about needing to know the intent of the rules. If the intention is to remove as much engine influence as possible then only very carefully selected databases can be used, but if the intention is only to prevent use of entirely engine game databases then just about anything other than 100% engine games will be legal. Of course there could be some other intention but we have no knowledge of that from the wording of that one small bit of the TOS.

How the content of my database could be assessed remotely is another matter. Should we all submit our databases for certification before using them? All I can say is I don't envy the poor chap who has to scrutinise every game in my 5,500,000+ game main DB. Whether or not that database breaks the rules I would imagine it would be next to impossible to determine remotely if I was using a database that does break the rules, especially since the approved cheat detection method discards the part of the game that the database would be used in.