[Retros] campaigning for linear email threads

[Andrew Buchanan]

Hi Guus,

We can save your poor 2003 problem which you have just cruelly
defenestrated!

If you feel my approach is liberal, loose or undeterministic, then I haven't
communicated successfully. I think my approach is rigid, but needs to be
fleshed out. Chess problems are "a movie of a book". There is some
adaptation from game space to problem space required, due to (1) absence of
knowledge of game history & (2) absence of human decision-making.
Conventions serve to cover these holes. But they are *not* temporary rules:
they are separate and need to be seen as such.

Why? Take the old e.p. convention, for example:
"An en-passant capture on the first move is permitted only if it can be
proved that the last move was the double step of the pawn which is to be
captured."
There are two ways this is different from a rule:
(1) The trigger depends upon "provability". 
(2) It's not asserting whether a move is legal, it's talking about whether
the move is *permitted*.

So the rules give us some sets of possible legal positions and possible
legal moves. We then use conventions to decide which moves are permitted/not
permitted. I am pretty sure Guus you are clear on the distinction between
permission & legality. But I am not sure you are so convinced of the
importance of keeping meta-rules involving statements of provability
distinct from the rules themselves. Be so convinced! This is not a new idea
- many paradoxes in logic are resolved by understanding the distinction
between rules & meta-rules.

The current phrasing of the castling convention is tiresome here,
introducing an undefined term "permissibility". A better phrasing would be:
"Castling is permitted unless it can be proved that the king or rook has
moved before." An analogous "provability" principle applies cleanly in
Article 15, covering first move.]

But the challenge is that DR (which is a rule) makes a statement about
provability. So we have two choices in how we partition into rules and
meta-rules. We could have something functional which says that anything
which talks about provability is a meta-rule. But I think that is too
complicated and requires case by case analysis. The fruitful way forward is
to use the natural division into rules and conventions which has been given
us. I don't think DR by itself when considered as a rule causes paradox.
It's the conflation with other conventions which causes difficulties. So
conventions=meta-rules.

Another example to back up this point. That players in compositions must act
to fulfil their stipulation (competing or co-operating) is a convention. If
this convention is another "temporary rule" then DR will cause every direct
pat, help pat or even draw study to be unsound! The source of pain is not DR
or the conventions, it's squashing together rules and meta-rules, when they
are clearly "trying" to be distinct. Note the same issue would still occur
if we regard *both* DR & conventions as meta-rules! So we really must make
DR a rule and conventions meta-rules.

I don't want to get into the detailed conventions for 3 rep & 50 move in
this long email. But these too should be viewed as meta-rules, while DR is
only a harmless rule which applies first, innocent of the conventions. So
Guus' wonderful problem is sound.

There are still many detailed questions to address, but this email is
intended to make one architectural argument. To what extent does it succeed?

Thanks so much,
Andrew

Gus wrote:

Ignoring the threefold repetition is indeed legal in a chess game, but
repetition problems are mainly based on the retro convention that 3R leads
to an automatic draw. You cannot claim that a convention automatically draws
in one phase (the solution), and does not apply in another phase (DR
evaluation). 
 
I just saw Andrews mail with his liberal take on the use of conventions. I
am glad he has a different view and I am sure we will not agree on this for
a long time to come. I can't however resist the temptation to make one
argument here. Andrew writes "... *then* use conventions as a last step to
resolve any residual uncertainty". It is important to observe that this is
not about uncertainties in rules but uncertainties on "retro-active-states".
The game rules are clear enough (applied with reasoning). The crucial point
is that composers create the state uncertainties on purpose in order to
invoke the use of appropriate conventions by the solver. As such they become
integral components of the composition and act there as temporary rules
within the problem space. The one thing however noone would not wish to
allow is that the "temporary rules" are different at different phases of the
solution. OK, they might vary prroblem to problem, but to allow
shapeshifting within the same composition would indeed amount to "Magic".
Accepting that one either uses the 3R convention in a problem or not at all
places a considerable burden on the composer. In particular, finding decent
stipulations becomes a difficult issue. With game rules, draw claims may
precede the actual repetition, and so should the diagram. This causes
confusion amongst solvers and composers. Draw(n) when? By the way, winning
is OK as well when the instruction is simply to "draw"..
 
I can't see how a loose and undeterministic approach will forward this
field. How can we ever hope to conquer the fairies if we cannot even provide
unambiguous clarity in the orthodox domain?, 
 
I think I just axed my 2003 problem. What I do for love...
 
Best wishes, Guus Rol.

-----Original Message-----
From: Retros [mailto:retros-bounces at janko.at] On Behalf Of Andrew Buchanan
Sent: 27 May 2014 23:41
To: 'The Retrograde Analysis Mailing List'
Subject: [Retros] Andrew answering

Hi Joost,

I agree with you, but what others are saying is that under the current
conventions, you can't ignore 3-fold repetition at all.

But in fact even with the current convention, the problem is still sound.
There is a crucial principle (unwritten, like most of the important stuff
apparently) which says that you should apply all the rules (including all
reasoning) exhaustively, and *then* use conventions as a last step to
resolve any residual uncertainty. So we should apply A1.3, which says yes,
this game is alive, indeed the last move prior to the diagram is the key
move in a direct #3. And then we look at how conventions apply. So I never
had any doubt this problem is sound.

If my impatience with the conventions seems strange, it's because I know
that our stupid situation is unavoidable. We are living in something like
the dark days of Magic the Gathering *before* they properly sorted the rules
out (around 1998, with the release of 6th edition rules). It was horrible.
And since Wizards of the Coast invented the stack, layering system,
templates, and all the other beautiful rules concepts which underlie modern
Magic, they've been going from strength to strength. Magic is *far* more
complicated than chess rules+conventions, but the rules are so solid they
don't even need judges for online matches. Check out
http://www.wizards.com/Magic/TCG/Article.aspx?x=magic/rules: both the Basic
Rulebook and the Comprehensive Rules.

What MtG has is a *corporation*, who can just define the rules and say
"that's that". We just have, bless us, a bunch of meandering nice people.
The only way out for us, I think, is if one (1) respected elder statesman
with time on his hands assembles a set of conventions properly founded in
logic. And then the rest of us will moan and complain when he proposes
these, but except for a few edits, what he will say will be accepted,
because it will be such a big step forward in common sense. That's the only
way we can grow our little hobby.

Magic has millions of players today - bigger even than when it started as a
fad in 1993. Why shouldn't retrograde chess problems have a few more
thousand serious enthusiasts than it does today, if we take away the biggest
barrier to entry & retention?

"Vote for Guus!" say I. :)

All the best,
Andrew.

-----Original Message-----
From: Retros [mailto:retros-bounces at janko.at] On Behalf Of Retros
Probleemblad
Sent: 27 May 2014 22:52
To: The Retrograde Analysis Mailing List
Subject: Re: [Retros] rights & ocassions / not answering Andrew anymore

On 05/27/2014 11:57 AM, Guus Rol wrote:
> Hi Olli,
> Yes, you got the idea! I am not sure about the precise position and 
> timing but basically DR cooks it if you aim for the position after Bf8.

I don't think so. DR uses article 5.1b ("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."). Since ignoring the 3-fold repetition is legal,
there's a legal continuation in which any colour can checkmate.

Joost.