[Retros] I'll say it again...cooked!

[Kevin Begley]

There's really two arguments here.

Christian Poisson seems to be claiming that extended castling is not an accepted rule for vertical cylinder. The vast majority of problemists, I dare say, do not share his view. 

I can understand his intent -- to preserve problems which once were sound -- but as I have expressed already, the best honor we can do for the original composer is to correct the problem for today's rules (and credit them for the original idea). 

The second argument is clearly expressed by TV's comments...


:    If a problem was sound under the conventions applicable at the time of

: its publication, I cannot see how a change in the conventions (as opposed to

: a cook found under the conventions then applicable) can render it 

: retroactively unsound.  Moreover, I would seriously entertain the alternative 

: that a seeming failure of perception was, in fact, a convention (whether or 

: not we  have evidence of its articulation). 


In fact, TV is arguing that problems can be "retroactively sound." TV's method for deterring soundness is the one which requires retrograde analysis of yesterday's rules. We do not apply old conventions to determine soundness of a chess problem, why should we retro-analyze old conventions to determine the soundness of a Vertical-Cylinder problem?

Consider Loyd's famous "Chesshire Cat" problem. The original is now well known to have been "incorrect" because the white player cannot have been on the move in the diagram. At the time of publication, it was not convention to apply retrograde analysis to a directmate. The incorrect problem has been made sound for today's conventions.

The same can be said of the original "indian" problem, which had two solutions (back then transpositions were not considered a separate solution, apparently). This problem has also been corrected.

If the problems are to be considered sound, based upon the time of their publication, why have they been corrected?  TV seems to be suggesting the solver should memorize the dates these changes took place, and solve based upon each problem's publication date. This is no way to define the rules of a game. No game's rules should now depend upon time and place of publication.

Either the problem is correct in today's rules, or it requires additional conditions (e.g., "Vertical Cylinder" plus "no extended castling").

It is nice, perhaps, if not common practice, to note (for the original author's sake) that the problem WAS sound at the time of publication -- that the correction was based upon a rule-change rather than a flaw in construction (show the original and state the rule change if possible). 

Whereas a "failure of perception" may have occurred in V-C, some chess problems composed prior to the en passant rule (which are unsound chess problems today), made no such error in perception (en passant is not the kind of rule change one could have easily perceived would occur). Nevertheless, this radical change did render several problems unsound.

Today's rules should be clearly defined, and any problem which is not sound under those rules should stipulate (by extra conditions) their differences, for the solver.

Kevin Begley
kevin_begley at hotmail.com

PS: I have seen it claimed, in vertical-cylinder problems, that a unit (such as a rook), which makes a move back onto its original square, can have moved in two directions, thus cooking the problem (with its very intent). I find this argument rather absurd, because the direction is of no consequence to the "origin-destination" which defines the move.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.pairlist.net/pipermail/retros/attachments/20020905/3984c069/attachment.htm>