No mating material = Draw

Einstein, of course there are numerous positions that are a draw according to the FIDE rules but do not appear on the list I gave above. Your position is one example.

The list given above was with the aim of implementing something simple that can catch most of the most obvious examples with an automatic material distribution criterion.

Everything else would very probably be either a new degree of programming effort, or simply would have to be left up to human arbiter decisions.

Since nearly all of these positions occur with less than eight peices, could a tablebase not be employed to make the determination?

Here is one that white would win in all variations:

the problem is there are tens of thousands of games per day, and we can't do manual arbitration. we need computerized rules. 

You mean a kind of tablebase which consists only of positions in which there is no legal sequence of moves that leads to mate?

Could be a good solution to the problem for most cases, but it seems to give only an incremental advantage over just using a simple material distribution table as given above ..., while the effort to create a reasonable part of such a database would seem to be rather big at first sight.

Well, the databases already exist -- it would just be a question of tying them into the decicion making process on timeout.

Too cool.

which requires more processing power than we can dedicate :(

we need SIMPLE rules. what is wrong with what i proposed?

K
K+N
K+B
K+N+N

all of it regardless of what YOU have - it only matters what the OTHER guy has when you time out. if he can't mate you reasonably, then it's a draw. 

Sure, so certainly the aim will be to computerize as much of the rules as reasonably feasible.

So one will probably choose to computerize all the material distributions named in the list in post #39 of legally inevitable draws, and choose not to automatize the draw of all the other positions.

Because, of course, one would never want an automated program to interfere and stop a game when a KNN vs K endgame is reached that is a forced mate on the next move. And one would certainly not allow it to become a strategy to just time out on the K side of this endgame in order to escape this mate in one threat with a computer-awarded draw after timeout.

This is why there will be some things that cannot be included in such a computerized solution. Automatized rules can be helpful oftentimes, but of course, they should not be allowed to destroy the normal course of a game of chess in ways such as mentioned in the examples given.

(Unless, of course, it should prove possible to obtain and make use of some sophisticated tablebase designed for this purpose. Using some live-engine assisted automatic decisions may be worth a thought, but would seem less efficient and probably even more difficult to implement than such a tablebase.)

Erik, if you know that the other guy has a knight, but you do not know what this guys has, then you are not able to call it a draw without violating FIDE rules in many, many cases.

Same for a bishop.

If he has KNN, you will violate FIDE rules in all cases if you call it a draw, except for extremely rare exceptions such as mentioned by Einstein in post #46.

The simplest thing we can give to you is a list as I have shown above, in post #39, that takes into account also the opponent's material.

Indeed a very important point is that K+2N can often force mate against K+P so you definitely can't award a draw if the K+P times out!

Furthermore, according to wikipedia the following position of K+2N against K+3P is winning for the knights

And I am not sure what you wish to mean by "mate reasonably", Erik. I would not want to have to experience going for a theoretical win such as

and be told it is a draw by some automatic machine that stops the game or awards a draw after my opponent times out.

I have an idea... play longer time controls with an increment

 It's nice to see this topic being discussed so thoroughly.

I do understand the theoretical downside to a K+P vs K+N game being deemed a draw when the side with the Pawn loses on time (which was my game in my original post).

But I am curious how many times this downside has occured in your games?
Do you all have numerous real examples from your own games to support the case that this theoretical downside is a realistic concern?

In my personal experience...
In my hundreds of thousands of blitz games in my life, I remember winning none with K+N vs K+P.
On the other hand, scoring a draw with the K+P vs K+N would not be considered uncommon in my games at all.

---------------------

we need SIMPLE rules. what is wrong with what i proposed?

K
K+N
K+B
K+N+N

all of it regardless of what YOU have - it only matters what the OTHER guy has when you time out. if he can't mate you reasonably, then it's a draw. 

---------------

I love it!

I recognize that this is not what USCF rules state.
But then again, I don't think there are many 3 0 games in the USCF where this situation becomes much more relevant.

It seems logical that sites that promote blitz chess should not necessarily blindly follow the rules that the USCF sets for games of slower time controls.
Instead, they should consider implementing an occasional rule that is in the spirit of the game of blitz and enhances the enjoyment of the game for the majority of players.

--cm 

CoranMoran, one of the great strengths of chess.com is that it is a great teaching site.

Automatically ruling your material distributions as a draw (or doing so after a timeout) would drastically imply that many chess positions that are winning would no longer be winning in chess.com games.

I would recommend against that because it would teach many, many newcomers to chess some strange rules, and it would make strong players feel taken aback about what they will see as a bug in implementing chess rules.

That is a good point.  :)  Black might deliberately time out to avoid the checkmate that is coming which seems unfair for white in that case.  I agree that finding the perfect set of rules for online chess is difficult.

I would not really call it difficult, as soon as one goes down from "perfect" to "reasonable and efficient". It seems to me that applying a material distribution criterion such as in #39 is easy enough to implement and covers most of the relevant cases in a fine and efficient way.

i totally understand your point ilmago, but the chances of that happening (a position where white can mate with just a knight, for example) are SOOOOO rare compared to the chances of someone "losing" on time when their opponent CAN'T force mate with a knight... is it worth is to inconvenience people 99% of the time for the 1% that would be more "correct"?

Erik, why don't you create a P4 bug, for tracking purposes.

Erik, do you really consider it to be acceptable to change the rules of chess and build a system that makes it impossible to play a normal chess game on the board in some cases? I would see a high danger in people starting to say: "On this server, it is not possible to play normal chess, because here and there, they have started implementing strange rules ...".