# [Retros] Retro solving competition Messigny

Mario Richter mri_two at t-online.de
Wed Jul 7 08:20:45 EDT 2004

Hello,

Pascal Wassong wrote:

> There was one special problem by François Labelle (like the one : at

> the 5th move black plays Rh1#, find the game), 5 SPGs by Christoph

> Fieberg, Joost de Heer, Satoshi Hashimoto, Görand Wicklund and Rustam

> Ubaidullaev, one classic retro by Tom Volet, and one classic retro

> with the Madrasi condition by myself.

...

> The were 8 problems to solve. Each problem gave 5 points if found.

> Here are the results of the retro solving competition :

> Place, name, points time

> 1. Michel Caillaud 38 1h59

> 2. Thierry Le Gleuher 35 2h00

> 3. Gerd Wilts 31 1h58

...

Regarding the results table it seems possible to
loose some points by giving incomplete(?) solutions.

So my question - especially with respect to the
SPGs and the "game determining final move problem"
by François Labelle that were used in that contest -
is:

What constitutes a correct AND complete solution
of such problems?

i.e. is it enough to give the moves that lead to the
final position resp. a game that finishes with the
given final move (correctness)?

Or has one also to prove that the solution found
is unique (completeness)?

To illustrate the above a little bit more, I assume
the task is to find the solution to the question:
"A game finished with 5. Ng3#. How did the game go?"

(This problem is published on François Labelle's
and referenced on E. Angelini's website, so I
think it was not the one that was used in the
contest.)

A solver's reasoning might go as follows:

The check might be
(1) a direct check by the knight or
(2) a discovered check.

so let's first investigate (1):

Black has 4 plies, the first of them not being a
king move, so the king can only reach the 5th rank,
and hence the king (being in check by the Ng3)
must be either on (1.1) f5 or on (1.2) h5.

Let's further assume the king is on f5.

Fastest path for a white knight to reach g3 from
it's initial square is Ng1-e2-g3, so let's try
1. e3 as the first white move.

For a king on f5 being mated by Ng3, all the
surrounding squares must be either blocked by black
pieces or attacked by white pieces.

So let's see:
e4 is attacked by Ng3
f4 is attacked by Pe3
e6 could be blocked by a black pawn.

The squares remaining are those of the g-file
and f6 and e5.
These squares can be attacked in parallel by a white
queen on g7.

Taking all this together, we get:
1.e3 e6 2.Qg4 Ke7 3.Ne2 Kf6 4.Qxg7+ Kf5 5.Ng3#

End of the hypothetical solver's reasoning.

So with only considering a small fraction of all
possibilities a solution is found.
But without considering all the other cases and
subcases one cannot be sure that this solution
is the only one.

Would mentioning this solution already give the
full 5 points or only a smaller amount (How much then)?

Thanks for any clarification!

mario