andrew at anselan.com
Sun Oct 6 05:20:10 EDT 2013
I didn't invent the term "synthetic game", but I like it here because the
term "proof game" is too overloaded.
One frequent user of the term is Jeff Coakley in his chess cafe column:
"Each column on proof games concludes with a synthetic game. Instead of
finding the move sequence that leads to a given position, the task is to
compose a game that ends with a particular move."
See in particular Synthetic Game #1 in
From: retros-bounces at janko.at [mailto:retros-bounces at janko.at] On Behalf Of
Sent: 06 October 2013 09:10
To: retros at janko.at
Subject: Re: [Retros] rstan
On 10/05/2013 10:40 AM, Andrew Buchanan wrote:
> I do have a shorter one, but the minimal solution is missing a capture
> that the longer solutions include. I suppose therefore it shouldn't count.
For the "shortest infinite game" task, I think that the problems should be
of the form
A * B^n * C (where n >= 0)
and the goal is to minimize |A| + |C|. Your record has
A = [d4 ... dxe5 ... Qxd7+ ... ]
B = [Qd6+ ... Qd7+ ... ]
C = [Qc6 ... Qc4 ...#]
There cannot be a capture in B.
If you want a new challenge, maybe try to find an example with |B| = 3?
> If any more of these are produced, I hope it's not too late to ditch
> the inappropriate & misleading name. These are nothing to do with proof
I'm not sure if "half proof game" is the best name, but these problems do
have something to do with proof games, in the sense that the goal is to find
a game starting from the initial position.
> I would use the term "half synthetic".
The name is usually what we have to find, e.g.
"mate in 2" means "find a mate in 2", and "proof game in 6.5 moves" means
"find a proof game in 6.5 moves".
Here, we're asked to find half of a game (because the other half is already
given). So saying "find a half proof game" doesn't sound that bad. Saying
"find a half synthetic" sounds strange.
> They have the feel of kriegspiel, too.
Yes. Also similar are Steno-Chess problems (
http://www.cetteadressecomportecinquantesignes.com/StenoC.htm ) and HAP
problems ( http://abrobecker.free.fr/chess/hap.htm ), which both also ask
for a game given partial information about the moves. These were discussed
on the mailing list in 2004 and 2007.
> On disambiguation, I think option c is the most natural & the best:
> "A fractional synthetic is obtained by taking the solution as written
> in algebraic notation and blanking the hidden plies."
Sounds good, thanks. I might be able to generate some "half proof games"
by computer, but first I wanted a precise definition.
> -----Original Message-----
> From:retros-bounces at janko.at [mailto:retros-bounces at janko.at] On
> Behalf Of Francois Labelle
> Sent: 05 October 2013 02:47
> To:retros at janko.at
> Subject: Re: [Retros] rstan
> On 10/02/2013 09:14 PM, Andrew Buchanan wrote:
>> Shorter again:
>> 1. d4 ... 2. dxe5 ... 3. Qxd7+ ... 4. Qd6+ ... 5. Qd7+ ...
>> 1004. Qc6 ... 1005. Qc4 ...#
> C+ again. Here's the solution with 2.0 moves added to the shortest line:
> 1. d4 e5 2. dxe5 Ba3 3. Qxd7+ Kf8 4. Qd6+ Ke8 5. Qd7+ Kf8 6. Qc6 Qg5 7.
> Qc4 Qxc1#
> On 10/02/2013 08:53 PM, Richard Stanley wrote:
>> One could also look at "half proof games" with various conditions.
>> For instance, does there exist such a game that ends in mate by a
>> pawn promoting to a knight? The moves of the side being mated are
> Interesting. We don't know the full potential of "half proof games"
> yet, but right now I worry that the genre isn't well-defined. The
> problem is what to do with the disambiguating moves in algebraic notation.
> Take for example the white moves
> 1. Nf3 ... 2. Nd4 ... 3. Na3 ... 4. Ndb5 ...
> We could take any of the following viewpoints:
> a) Half proof games never use disambiguation. In the example we would
> write "4. Nb5" and determining which knight moved would be part of the
> b) Half proof games always use disambiguation. In the example we would
> write "4. Ndb5" even if Black captured Na3 in move 3, so as not to
> give information about whether Na3 was captured or not. In other
> words, all the white moves are crystal clear as if we used long-hand
> notation everywhere (1. Ng1-f3 ... 2. Nf3-d4 ... 3. Nb1-a3 ... 4. Nd4-b5
> c) Half proof games use disambiguation according to algebraic notation
> rules (
> ). In the example, writing "4. Ndb5" would imply that Black didn't
> capture Na3, and writing "4. Nb5" would imply that Black captured
> Na3 or Nd4. In other words, a half proof game is obtained by taking
> the solution as written in algebraic notation and blanking the hidden
> I wonder which viewpoint people think is best. So far the problems
> people have posted aren't affected by this, but if the genre takes
> off, someone is going to test those boundaries someday...
> Retros mailing list
> Retros at janko.at
> Retros mailing list
> Retros at janko.at
Retros mailing list
Retros at janko.at
More information about the Retros