Hardest Mate in 4 of All Time

And I would be willing to bet that they will all be given the same answer (at some point) as to why their proposal fails.

There are times when you should leave it to people to find the answers within themselves.

I withdraw my previous comment. I mistakenly thought TheMoonwalker had posted previously in this thread.

When I comment, even if it's just a small comment, I draw so much attention that people believe I'm the only person talking. 

Now OP pls tell me that I've found the right answer..

@Aristokle:

If puzzle contains a retro-feature, i.e. the diagram contains hidden information from retroanalysis, it should be described in the stipulation since retro-problems belong to different category of problems. The OP decided to have fun instead and told nothing. It is fine, but he should accept a criticism for this behaviour. Indeed, the OP's diagram looks fishy at the first sight: missing unmoved bishop c1 and "redundant" rook h8 to help black king escape the primitive checkmate b6-b7-b8=Q#.

The retro-strategy convention is absurd. Future can't change the past. From billions of possible proof games, only one is real - that one written on scoresheets. Nobody can ever change that. Imagine e.g. this: President of Iran declares Holocaust didn't happen. According to retro-strategy logic, he just saved millions of lives under Nazi jurisdiction during WWII and should be given thanks for this, because Holocaust ceased to exist in real history. Good work, huh?

Look again at the OP's diagram. You play white, you may check your scoresheet, you know both sides castling rights, spectators don't. Now, you announce you see checkmate in 4, there is no sooner checkmate, and you make a bet with people around about it. My questions are:

A) If black can't castle, why would you play 1. Qc4 then, instead of immediate 1. b7? How do you prolong black's defence if spectators suggest a shorter 1. b7?

B) If black can castle, how do you deliver checkmate in 4 in order to win that bet?

The OP has already posted the answer, in comment #72.

Aw.. oh I see Im wrong cause in my line black can ignore rook and just castle instead.. then it's mate in 5, not 4 :P

Well, actually you're exampled meant to show absurdity makes it make sense to me...

If the president of Iran declare the holocaust didn't happen, and he's telling the truth, then it couldn't have happened.

Do you suggest that moves on the chess board can lie?

I am annoyed that it's a retro problem but I think it's pretty creative.

If you play 1: b7, black shows you a proof game showing that he has the right to castle, and castles.

If you play Qc4+ and castle, black cannot give you a proof game that shows he can castle. Therefore, he can no longer castle.

@A: The conventions for all types of compositions are in the same Codex. There is no requirement to specify that a problem is of the retro-type since this is considered an aspect of a problem and not a separate category. It is up to the solver to detect when it is around. Even if you believed that problem #1 is just an ordinary direct mate then you could still have seen that there is an issue with the castling rights. The diagram information is simply insufficient to solve the problem. That should have put you on the scent of the "retro-"conventions in the Codex.

Note that you already invoke basic "retro-"conventions when solving problems. Every time you castle (or consider castling), every time you reject an e.p.-capture in a diagram, you do so on the authority of Codex-conventions. The PRA/RS stuff is in the same Codex article to fill information holes left behind by the basic conventions. It is just a touch more complicated.

@B: It is OK to believe that RS is insane. RS cannot be justified with the common inference rules for directed graphs (chess). Humans however use RS-logic all the time. It is not about the future changing the past, but about assigning truth values to unknown pasts based on future observations. I will say no more about it. It will be in my book "The theory of everything (unimportant)".

@Arisktotle: As you well know, and as I am coming to appreciate, a composition is just a position, not a snapshot in the history of an actual chess game. As such, it has no past, but rather a collection of possible pasts. The position itself, and the moves that ensue from it, allow one to draw conclusions about those possible pasts. However, I don't think you will be able to convince Polar_Bear to agree with that understanding of what a composition is.

If so, I forgive him for it, for introducing me to such an interesting problem.

I am not so sure that Polar_Bear's objection is as you state. It is obvious that he can't maintain a blanket rejection of histories while applying the conventions for castling and e.p. to solve problems. His main issue, I think, is with the lack of information on precisely which retro-rules, -assumptions or -conventions apply to a particular diagram or problem type. I hope I have shone some light on that.

@Polar_Bear Addendum to #114:

This is a pre-reply to your next question. Of course there is a category of truly retro-analytical problems like proof games and 'resolve the position' types. However, this category is transparent to the Codex as much as the category of 'ice cream flavors' since it doesn't invoke any "retro-"conventions. In the absence of forward play, everything about the histories needs to be analytically proved including castling rights.

I have named the attribute that makes a diagram susceptible to "retro-"conventions retro-active. The term is intentionally similar to radioactive. Both refer to "futures contaminated by past events". Both have waste products it is hard to get rid of, see the intricacy of the conventions and the discussions in this thread . The attribute might be included In diagram stipulations e.g. "retro-active direct mate", "retro-active helpmate" or "retro-active endgame". The Codex doesn't requires it since it considers every ordinary problem or endgame as potentially retro-active. So keep your Geiger counter up close at all times!

That's right. The "possible pasts" I referred to do not include any illegal moves.

Look here:

http://www.chessgames.com/perl/chessgame?gid=1007158

Black wanted to play 22. ... 0-0-0, but white showed scoresheet with real proof game that Ra8 had moved, so black was forced to move with king (22. ... Kf8) since he touched it. Black's attempt to castle didn't change game history and castling remained illegal. Reti wasn't allowed to castle, despite the fact nobody brought retro-analytical proof he couldn't, except scoresheet.

This is how order of things goes. So, if in the OP's puzzle game history says black can't castle, puzzle has shorter solution and is busted. If black can castle, white can't fulfil stipulation and puzzle is busted too. This is that simple. Unless you suffer with magical thinking that your moves can change the past.

It seems I was right.

You missed the point of the Codex-conventions. The past is unknown (also to all OPs) and that legitimizes the addition of arbitrating conventions. I am sure you have castled during problem solutions even though you couldn't know that was legal from the past. If you believe it is wrong for the OP to let you guess in #1, it is also wrong for you to have played a castling move in any puzzle on chess.com. Nobody ever specified that castling was permitted - except the castling convention in the Codex.

If you are serious about your objections, you should address the relevant arguments against them.

Yep, but I doubt it speaks in his favor.

The past may be unknown, but supposedly exists. Do you realize that puzzle is incorrect in both possible cases? If the past doesn't exist, it means the position has been set up arbitrarily and game history is irrelevant to castling rights. You can study the past, but not create it. OK?