What's the quickest possible draw?

I don't know if this will help speed it up or not, but another way to get rid of the pawns is to promote them to bishops of the same color. This position is also insufficient mating material:

Now obviously this position would take many more moves to construct lol, but the point is if it takes more time to capture the last pawn than to just promote it to a bishop, that could be a possibility.

How about draw by repetition?

Well done in getting it down to 16.5 moves!

The record, though, seems to be 16.0 which was found using software, as part of a different project where problemists/programmers aim to uncover unique proof games that leave just a handful of pieces on the board. They weren't specifically aiming for dead positions, but among the sound positions found with a single precise solution, some were dead positions. In the following case, the final position has 3 pieces left, but at move 16.0 prior to the last capture, the position with same-coloured bishops is already dead.

@Rocky64

An irrelevant point in the context of the thread, but relevant to the proof games you refer to. Move 17 is not part of the game. A dead position immediately terminates the game.

5.2.2 The game is drawn when a position has arisen in which neither player can checkmate the opponent’s king with any series of legal moves. The game is said to end in a ‘dead position’. This immediately ends the game, provided that the move producing the position was in accordance with Article 3 and Articles 4.2 – 4.7.

So the software didn't find a game reaching KBvK. The final position in the game has four men.

(16...Bg3 17.Bxg3 Kxd7 would have done the job with an extra ply. But not unique; 16...Ba7 does just as well.)

Nice!

Along this line, I now found several ways with 32, but so far still failed to get it to 31. Here is what I was trying to accomplish, since it's actually possible to keep 5 pieces on the board:

There is a lot of confusion here. "Chess" either is a "forced draw" or a "forced win". And that is the only answer to your question and it applies right on move 1.

I had a discussion with a game theoretician a while ago and I said the game players often mean something else when they say "forced". Game players commonly mean "controlled win" or "controlled draw" to indicate that the result is within their ability to calculate or is very simple.

What you mean is that the position you call a "forced draw" is "dead" by the rules of the game (article 5.2.2). K+N vs K+N is not dead and therefore not a "forced draw". However, the designation "dead" has no specified handles. When a position is dead, it can be because stalemate is unavoidable, or the 5-move repetition draw is unavoidable, or the 75-move "no progress" line is unavoidable. The endless roaming of pieces is fictional since it is always aborted by one of the 3 "drawers" given.

Interestingly you can exclude the 3- and 5-fold repetitions, and the 50 / 75 move no progress rules and the stalemates from your search but not the "dead" - rule as then there is nothing left to attach your "forced draw" to. You would have to introduce a new rule in chess to define an acceptable "forced draw".

Also note that "dead" is a state in itself and not the same as the state it "anticipates". So a position will be dead if a "stalemate is unavoidable" not because it is "stalemate". It is easy to construct a short proof game which draws in a "dead" state, because a 5-fold repetition is unavoidable! Btw, "@jetoba" (FIDE arbiter) will drop a query with FIDE on these issues which have not even been noticed there!

Finally, for handling rule issues you need to state where you work under FIDE rules and where under Composition rules. The Composition rules are more complete - though there are still lots of issues ha, ha - and sometimes different. For instance, it demands automatic draws on 3 repetitions or 50-moves no progress.

In other words, you are in a wasps nest!

Astute insight, indeed! Have you checked how many games I've played?

Forget about 'forced draw,' Arisktotle. Just find me a position with a minimum of 25 pieces on the board, both players can still move, and it's impossible for either to deliver mate.

Sorry. And of course no termination from a forced 3-5 move repetition.

There are no positions where both players can move.

Of these, which would you exclude if the 25 piece requirement were relaxed?

Here is an example. Game played with FIDE rules. Position is dead because black will cause a 5-fold repetition with his next move, not because one is there right now!

.

If played under FIDE competition rules. Not dead under FIDE basic rules (as is e.g. second example in my post above).

Nobody said "basic". These are the common competition rules. In fact, considering that none of this is in the game domain, the standard rules are the composition rules, not the basic rules.

Both. Though if you can find a forced stalemate position in which both players can now still move, with 25 on board, I'll still be obliged.

I want a position which isn't a forced stalemate, and in which the game is terminated even if repetition rules didn't exist, simply because none can possibly mate the other.

Something like this, but with a minimum of 25 on board.

I made a mistake, MARattigan, you're 2nd position is good.

Proof games are compositions, not practical games, so composition rules trump FIDE rules.

The WFCC Codex states:

Article 17A – Dead Position Rule. Unless expressly stipulated, the rule of dead position does not apply to the solution of chess compositions except for retro-problems.

https://www.wfcc.ch/rules/codex/

Although proof games are usually classified under retro-problems, it's obvious to problemists – who understand the point of these unique proof games – that the Dead Position rule is simply irrelevant in such cases and rightly ignored as in most compositions.

I stand corrected regarding the appropriate rule set.

I'm not so convinced by your statement that the dead position rule is irrelevant in such cases. It conceivably alters the acceptable moves in a solution and possibly the number of moves in the shortest solution, which would seem to qualify as relevance.

(Do you know where the WFCC defines "retro-problem" by the way? I can't find it.)