[Retros] rights & ocassions /answering Andrew

Kevin Begley kevinjbegley at gmail.com
Wed May 21 17:41:35 EDT 2014


Thank you very much for a fascinating read!
I certainly did not appreciate the complexity of the matter, and I really
appreciate your taking the time to detail these issues.

I hope this is not a dumb question, but I'll ask anyway:
Have you ever written a book on Retros?  Because, I think I'd really love
to read it -- especially if you help lead readers through the more complex


On Wed, May 21, 2014 at 4:00 AM, Guus Rol <grol33 at gmail.com> wrote:

> Hi Kevin,
> I agree, the "static" metarule should apply to all rights evaluations,
> including e.p., castling and all the attributes we don't know about yet in
> the fairy types that are yet to be invented.
> 3R is the most complex of retro-active properties in the Orthodox field,
> since an enormous amount of information bits may be required to record all
> the possible diagrams + plus their counters which may be relevant
> for future play. 50M is much easier, one counter tells all there is to know.
> Fortunately, this is where the retro-logics come into play. The
> (chosen) logic allows us to make simplified assumptions about the "real
> world" (an imagined real world) . Nobody uses the "consequent
> retro-variants" from one of my previous posts since handling all histories
> is practically undoable..In fact, almost all diagrams have a history where
> the diagram position is an  automatic draw. Zero options left to reach a
> goal. There is however nothing wrong with the basic path splitter
> of prA-logic - only maximized combinations of preferred retro-active
> conditions - when solving retro-problems. With regard to 50M and 3R it
> means we preferably assume the last move was a capture or Pawn move. Thus
> we do not run the risk of colliding into these rules by surprise. If such
> an assumptions is not legal we minimize as close to this assumption as
> possible.
> For the 3R case things can remain complicated as one or more full diagrams
> may still need to be included in the "positional information". In actual
> retro-problems the situation simplifies further since we can eliminate all
> but one past(s) for which the problem solution is the same. Only when a
> particular past collides with a particular solution, one is required to
> generate a second branch and a second solution. This is how we arrive at
> the partial retro-variants. You can e.g. see how this works in my own
> problem R309 in Probleemblad (I think 2007). Though it is not to be solved
> by retro-vartiants, the solution shows how and why this could have been a
> prA-type diagram with relevant 3R attributes. The reason I object to prA is
> not for its principles but because prA in de Codex was reduced to
> handling cases instead of giving it status as a generic logic.
> Justification. Why is it justified to reduce the history tree a priori in
> prA/retro-variant type problems? Well, mainly because it is in line with
> basic convention pholosiophy. By allowing you a license to castle, it says
> you need not solve the diagram history without castling rights. So why
> require it without necessity when 2 or more retro-active attributes are
> involved? The second point is the concept that retro-variants are not
> merely created to reflect different histories, but to demonstrate different
> forward solutions. Every solution corresponds with a whole class of diagram
> histories, there is no point in listing them individually. In some cases
> the sharing of solutions between different histories will be considered a
> defect but such judgement is on a higher level than discussed here.
> Best wishes, Guus Rol.
> On Wed, May 21, 2014 at 1:54 AM, Kevin Begley <kevinjbegley at gmail.com>wrote:
>> Thanks Guus,
>> I agree that units awaiting rebirth should be treated as a "static" and
>> independent feature of the position (read: as bits of information necessary
>> to characterize a position), regardless whether those rebirths can be
>> executed.
>> But, the same should be said of castling rights (or licenses) --
>> regardless whether castling can be executed.
>> I also agree that you can derive the minimal set of information (required
>> to characterize any position) from a simulated branching analysis of all
>> possible games, but once established, no part of that independent "static"
>> information can be excluded from the comparison, in establishing 3R.
>> You can not do branch-analysis on the fly, from any diagram position, to
>> create a new minimal set of information criteria (the minimal set of
>> information must be capable of describing every possible position).
>>  In this informational approach, repetition implies nothing more than an
>> exact match, spanning every single bit of information, from this minimal
>> set necessary to describe any position.
>> Best,
>>  Kevin.
> _______________________________________________
> Retros mailing list
> Retros at janko.at
> http://www.pairlist.net/mailman/listinfo/retros
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://one.pairlist.net/pipermail/retros/attachments/20140521/71a3923b/attachment.html>

More information about the Retros mailing list