[Retros] Favour/En Passant
andrew at anselan.com
Mon Jun 16 06:58:13 EDT 2014
I disagree again with your main point.
First, I do agree that Jonathan’s diagram is an excellent case. But I think it’s handled adequately by the existing conventions. It is not an issue nor a joke.
Before the recent changes to the conventions driven through by Werner Keym, under the en passant convention the diagram would have no *permitted* mating move. That’s fine. The sky doesn’t fall apart. Even if there was no other legal move for White, this doesn’t put White in stalemate. Stalemate is the result of having no legal moves. It is *not* the result of having no permitted moves.
It’s not a crisis: the sky doesn’t fall. We just admit the fact that in a few diagrams, it is not possible to determine whether certain en passants are allowable, and indeed whether the game is over or not. Those are interesting diagrams, and it’s certainly not worth twisting everything in order to ensure that some en passant must be allowed. That would be horrible.
Under the PRA convention, which now applies by default, things are even more straightforward. The problem splits into two parts, according to possible histories. Each history allows one en passant, so the problem is solved. I believe that Werner Keym would say that the problem has 1 solution, in two parts. I don’t particularly like PRA in this context, because it excludes any history in which neither e.p. is allowed. That just makes things too easy: most of Jonathan’s work is about showing that Black’s last move must have been one of the double hops. This is rendered irrelevant by PRA. However, PRA is not the main point.
I think our main disagreement is coming clearer. Once again you are mixing together rules and conventions. My position remains that just because the conventions don’t *permit* us to play a certain move should have no impact at the level of the rules themselves.
I made the next point in an earlier email, but I don’t think I got it across to everyone so I will say it again.
If rules and conventions operate at the same level as you propose, then *every* help pat is unsound, because the players are constrained by convention (i.e.: the definition of mandated player behaviour in a help pat) to work with one another to reach stalemate. Alternate lines of play which do not end in a stalemate are irrelevant, because the players are not permitted to play moves that diverge from the solution. So A1.3 kicks in right at the start. Similarly, *every* direct pat is unsound. White is bound by convention to eschew any path that can avoid a stalemate, and despite Black’s best efforts, the composition must end in a pat. Hence A1.3 will kill the solution right from the start.
Now the issue here, I submit, is not A1.3, but the mixing together of rules and conventions. Please let’s distinguish the rules from the conventions, and say the rules are about legality, while the conventions constrain which legal and possibly-legal moves are permitted, then we do not get in the horrible confused state that you are proposing.
And I don’t even believe that your planned path makes it easier to scale to fairy compositions. I think the idea of rules (including fairy rules) addressing legality, and conventions (including perhaps fairy conventions) then determining permissibility, is a bedrock upon which we can build a solid artifice.
From: Retros [mailto:retros-bounces at janko.at] On Behalf Of Guus Rol
Sent: 16 June 2014 17:33
To: The Retrograde Analysis Mailing List
Subject: Re: [Retros] Favour/En Passant
Excellent case and a real issue. There are many similar situations, particularly in fairy land like with the "fuddled men" in Turnbulls infamous article. I had preseved the treatment of these cases for an "advanced topics" post but I can outline it here. You could have made things a little worse by making sure that white was "stalemated" apart from playing the e.p. moves. Deciding that white is stalemated would be illegal as there would exist no proof game leading to stalemate. Clearly some e.p. move must therefore be permitted.
The reduction principle on the DGCs is based on the premise that some preferred option for play remains but such is not always the case. Besides that, there is the possibility of a "group right", a right that cannot be proved for each individual member of the group but van be proved to exist somewhere in the group. The handling of the cases occurs on a higher level than the handling of the DGC-set / Game set.
The natural approach is the temporary promotion of all "secondary rights" to "primary rights" (I am not sure about this terminology yet but you know what I mean) in a "rights group" allowing each one the be executed as such - i.e. as if it were a right to castle. After this promotion, the reduction from DGC to Game resumes with the modified "rights".
One issue remains and is reflected by your example. Do you wish to allow "promotion" only when no other playing option remains or do you wish to allow it whenever a group right exists - as in your example? There is an aspect of personal taste in this choice, but also one of best workability. Having contemplated this for a while on the basis of fairy forms, my gut feeling is that it is best to stick with the first choice. Which means that white cannot play e.p. in the example you presented.but he could if he were stalemated (or mated) otherwise. But you are entitled to disagree since the choice steps outside the necessities of a a sound and consistent decisioning system.
Note: On top of the retro-decisionig-tree is the 1st command: There must always be a proof game.
Best wishes, Guus Rol.
On Mon, Jun 16, 2014 at 10:22 AM, A J Mestel <ajm8 at maths.cam.ac.uk> wrote:
Now that I'm connected again, I'll repost something I wrote a month ago.
It feels a bit like a poor joke, which had some point at the time, but loses everything in the re-telling, but still, here it is/was.
Date: Fri, 6 Jun 2014 11:04:56 +0100 (BST)
From: A J Mestel <ajm8 at maths.cam.ac.uk>
To: The Retrograde Analysis Mailing List <retros at janko.at>
Subject: Re: En passant
Did this ever get posted in the list? I never saw it, and got no replies.
It's a bit dated now, but someone said that en passant was only legal if you could prove what the previous move had to be etc. Needs to be a bit more precise.
On Wed, 28 May 2014, A J Mestel wrote:
Someone must have done this before, but consider:
W: Kc5 Rd8 Bc8 Nc6 a5 a6 e5 e6
B: Kc7 Bb8 b5 d5
Can White mate in one? Not according to the definition I read here a few
Retros mailing list
Retros at janko.at
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Retros