[Retros] rights & ocassions /answering Andrew

Kevin Begley kevinjbegley at gmail.com
Fri May 16 15:11:47 EDT 2014


ps:  Guus, I wonder if you might be conveying the following... ??

To full convey the actual position is not a matter of information (4-bits,
6-bits, or otherwise), but a matter of branching (some inherently minimal
set of possible alternatives, emerging from the vertex of a graph/tree,
which can span both directions: forward and retrograde).

If that's what you are saying, it is deeply fascinating -- I would never
have believed such a statement could emerge from a discussion of chess.
In fact, it seems almost a "many-worlds" interpretation of information
theory.

When my head stops spinning, I'll need some time to fathom this.

Kevin.






On Fri, May 16, 2014 at 11:56 AM, Kevin Begley <kevinjbegley at gmail.com>wrote:

> Hi Guus,
>
> I appreciate your very thoughtful reply, even though, in places, I found
> it rather vague, and difficult to follow -- at least that's my first read;
> I plan to reread this a few times, to fully appreciate your meaning (I may
> need to reply again, at some later point).
>
> I had the same feeling when I first read Kmoch's "Pawn Power in Chess."
> I could sense it was imparting something significant, and could find no
> specific fault with the presentation; yet, it was a struggle to fully
> digest the implications.
>
> For example, I don't yet appreciate the motivation for (nor the value
> contained behind) your defined term, "license to castle."
> I remember having the same reaction to Kmoch's many terms (e.g.,
> dispersion, distortion, interspan, ... etc etc etc).
>
> At least I did learn that I would very much like to read more from you, on
> these matters.
> And, I'd like to learn more.
>
> Kevin
>
>
>
>
>
> On Fri, May 16, 2014 at 3:21 AM, Guus Rol <grol33 at gmail.com> wrote:
>
>> Hi Kevin,
>>
>> You may have norticed my posts are not very short, even when attempting
>> to address a single issue. I'm afraid I cannot answer all the points
>> related to your message in a single post of decent length. I will therefore
>> restrict myself to one paragraph from of my own retro-theory and paradigm.
>> There are no isolated retro-issues, everything should in the end, be
>> coherent, unambiguous and programmable into one computeralgorithm.
>>
>> The "right to castle" in the conventions is not the same as the "castling
>> right" in the laws. Some years ago I introduced the designation "castling
>> license" (or permit) to replace the "right to castle" in the conventions. A
>> castling license is uncertain and its fate depends on the retro-logic
>> involved and the line of play. It is fair to state that all
>> retro-conventions are a service required to connect the chess laws to the
>> retro-logics. Retro-conventions should be limited to treating single
>> retro-active properties and counters. Plus the properties that are placed
>> in a dependency relationship by the laws such as "repetition and castling
>> right". In contrast, the retro logics treat all interference between
>> different instances of retro-active properties. Thus, the primary
>> components of every retro-active chess problem (orthodox or fairy) are: the
>> diagram, the FIDE laws, the applicable retro-convention, the chosen retro
>> logic and the stipulation.
>>
>> An example on how this relates to the "positonal properties" in your post:
>> Let's say "consequent retro variants" - in line with an article by
>> Alexander George - constitutes the requirement that a problem solution
>> branch must be presented for every possible game history. This is a retro
>> logic and it is easy to see that with it, not a single retro-convention is
>> required. The logic itself is "big enough" to handle every chess
>> diagram.as is. In softer forms such as prA (partial retro analysis) you
>> need not deal with uncertainties either since every branch comes with
>> preset conditions, e.g. B1 short white castling right, not long, B2 long
>> white castling right, not short. These are not licenses but actual
>> conditions within their branch. You do need the conventions though to
>> establish which branches must be included in the solution. Those are the
>> ones with the "maximized combinations of preferred retro-active
>> attributes". Things get (seemingly) messy when you start using
>> retro-strategy or a posterior logics. A considerable part of the complexity
>> in the examples discussed in this thread comes from the mix of
>> "conventions" with "retro-strategy". Retro-strategy becomes more digestable
>> once you understand its premises and "justification scenario".
>>  The "counters" are relatively easy to handle within retro-strategy
>> logic but can raise considerable issues in all the retro-variant types such
>> as prA. To ignore them violates the basic premise and justification that
>> "relevant retro-active conditions are known inside every branch" but
>> including them may spawn a number of unrewarding variations. Usually a
>> different solution is expected for every prA branch. This, plus the
>> additional complications coming from fairy types, is the reason that I
>> consider "retro-strategy" (as in post factum) the most appropriate default
>> logic for all retro-problems. The current conventions have attempted to
>> draw the logics inside the conventional space but this can never work. In
>> the long run the retro-logics will be reinstated as an independent
>> component and dimension in the retro object space.
>>
>> Not nearly an answer to all issues as I said, but a start.
>>
>> Guus Rol.
>>
>>
>>
>> On Thu, May 15, 2014 at 7:58 PM, Kevin Begley <kevinjbegley at gmail.com>wrote:
>>
>>> Hi Guus,
>>>
>>> Thank you for the clarification.
>>> I fully agree that all assumptions should be explicitly stated,
>>> somewhere in the Codex.
>>>
>>> However, there seems to be some confusion concerning the state of a
>>> position.
>>>
>>> An orthodox chess position may be reduced to the following description:
>>> 1) the diagram (board orientation, and placement of all pieces upon it),
>>> 2) the set of all castling rights,
>>> 3) the set of en passant rights, and (the one everybody forgets),
>>> 4) the dead-reckoning consideration (if no help-win exists, however you
>>> define win, the game is over).
>>>
>>> The above information is both necessary and sufficient to determine the
>>> set of all legal moves, in any position; therefore, the above information
>>> forms a basis for the position.
>>>
>>> Beyond this basis (the state of a position), there also exists some
>>> attributes of a position...
>>> For example, the counters (50-move counter, and 3-move repetition
>>> counter) constitute an attribute of the position.
>>> Alteration of these counters does not impact the the basis (or state) of
>>> the position -- the attributes have no impact upon the set of legal move
>>> possibilities in said position; they only have an impact upon the resulting
>>> game outcome.
>>>
>>> It is important to separate the aims and goals of a given chess game (or
>>> problem), from the rules of legal movement, for said game.
>>> And, the people who write the rules/Codex should know how to express
>>> these rules from a fairy chess perspective (accounting for varying
>>> stipulations, aims, goals, etc)...
>>>
>>> With regard to the second point (castling rights), just consider how
>>> orthodox chess skews the proper understanding:  in orthodox chess, castling
>>> rights are reduced to a set of four information bits:
>>> 1) white has K-side castling rights?
>>> 2) white has Q-side castling rights?
>>> 3) black has K-side castling rights?
>>> 4) black has Q-side castling rights?
>>> (each question can be answered by YES/NO, with the default being YES,
>>> unless spoiled by retro-analysis).
>>>
>>> But, the people who write the Orthodox Codex mistakenly believe that
>>> this information is fundamental to a position, when it absolutely is not!
>>> The programmers who develop a more encompassing representation (e.g.,
>>> including Circe/Anticirce forms, wherein castling rights may be renewed
>>> when units are reborn onto their game-array squares) must all appreciate
>>> that the above 4-bits are a fiction; actually, 6-bits of information are
>>> necessary to completely characterize castling rights in a position:
>>> 1) white King has castling rights?
>>> 2) white K-side Rook has castling rights?
>>> 3) white Q-side Rook has castling rights?
>>>  4) black King has castling rights?
>>> 5) black K-side Rook has castling rights?
>>> 6) black Q-side Rook has castling rights?
>>>
>>>  This example is just one reason (of many) that I've had to constantly
>>> remind the orthocentric composers to return to the fundamental truths (and
>>> establish sound definitions).
>>> It is why I have advocated for a Fairy Codex, for years -- not only
>>> improve the prospects for fairy chess, but also to clarify orthodox.
>>> You really don't understand the rules of a language, until you encounter
>>> the rules of other languages -- so it is with the rules of chess.
>>>
>>>
>>>
>>>
>>> On Thu, May 15, 2014 at 3:23 AM, Guus Rol <grol33 at gmail.com> wrote:
>>>
>>>> Hi Kevin,
>>>>
>>>> The concept I am referring to is that repetitions in chess games which
>>>> do not genearate positions with the same options for the players (as in the
>>>> example given) can still be considered "the same" by the FIDE laws. In our
>>>> retro-conventions we are thereby justified doing the same thing. We may
>>>> e.g. consider there is still a future possibility to castle - in judging
>>>> the sameness of a position - even when we can't actually get there because
>>>> of some automatic drawing convention interfering on the way. No two things
>>>> in the universe are ever really the same and so it depends on the
>>>> properties you wish to include in the comparison.
>>>>
>>>> On a more generalized level I think it is wise to postulate that
>>>> no administrative condition (50M and repetitions, automatic and claimed)
>>>> can be interpreted against another administrative rule - or a different
>>>> instance of the same rule. They are "transparent" to one another. Example:
>>>> you can't say that a position is not the same because we are now 47 moves
>>>> away from the 50M mark instead of 49 moves. And you can't say a position is
>>>> different because one move gives an automatic draw on the second instance
>>>> and didn't on the first. Just to show that this issue is not limited to the
>>>> subject of "future castling rights". The latter is merely a complex case.
>>>>
>>>> You make a good point about the "implicit" assumption that a diagram
>>>> position occurs for the first time. IMHO this is an assumption which
>>>> requires an explicit retro-convention. The reason is that the assumption
>>>> can interfere with other conventional assumptions such as "castling right"
>>>> and requires resolution by a retro-logic. This has been overlooked because
>>>> the assumption was just a little bit too implicit. See my second post
>>>> yesterday which is also on that subject.
>>>>
>>>> Guus Rol..
>>>>
>>>>  On Wed, May 14, 2014 at 10:30 PM, Kevin Begley <kevinjbegley at gmail.com
>>>> > wrote:
>>>>
>>>>> ps: perhaps we should encourage a retro competition (especially
>>>>> including fairies), wherein the stipulation asks for 3-fold-repetition in
>>>>> n-moves (for n < 3, which would obviously imply retro-content)
>>>>>
>>>>>
>>>>> On Wed, May 14, 2014 at 1:26 PM, Kevin Begley <kevinjbegley at gmail.com>wrote:
>>>>>
>>>>>> Guus,
>>>>>>
>>>>>> > "...we are left with the question on how to harmonize the
>>>>>> original FIDE concept with the addition of automatic draws in the world of
>>>>>> composition - still not considering castlings at all. I would suggest that
>>>>>> to "continue to ignore possible terminations" is most in line with the FIDE
>>>>>> concept. Yes, some repetitions do  change options for the players but so
>>>>>> does the original FIDE law in real chess games..."
>>>>>>
>>>>>> This issue has already been harmonized, by the implicit presumption
>>>>>> that such repetition is not inherent in the diagram, unless it can be
>>>>>> proved by retrograde analysis.
>>>>>> Several retro composers have even attempted to make good use of this
>>>>>> presumption (most notably in studies wherein the solver must make use of
>>>>>> this to either draw, or avoid draw); unfortunately, I can provide no good
>>>>>> example (the concept is obviously very difficult to achieve, which might
>>>>>> explain my having no recollection of an entirely successful rendition).
>>>>>>
>>>>>> Perhaps I am misunderstanding you, because I would have expected that
>>>>>> this presumption was inherently obvious (otherwise, it would cast suspicion
>>>>>> upon almost every #2 composition).
>>>>>>
>>>>>> Kevin.
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Wed, May 14, 2014 at 12:30 PM, Guus Rol <grol33 at gmail.com> wrote:
>>>>>>
>>>>>>> Starting with a puzzle followed by an introduction, the solution and
>>>>>>> an explanation.
>>>>>>>
>>>>>>> Make the followiing changes from PAS (initial game array): Remove
>>>>>>> Nb8, move Pe2 to e4, move Bf1 to h5 and  Ng8 to f6. This is an SPG in 4.0
>>>>>>> moves but it is not unique. How many legal  forward moves do you need to
>>>>>>> play forward at least to make sure that the first 4 moves in the game are
>>>>>>> unique (then)?
>>>>>>>
>>>>>>> This puzzle shows many different aspects of the issues discussed. It
>>>>>>> also adds a new one: retro-logics. Whenever you do anything in
>>>>>>> retro-solving which moves both froward and backward in time, you need a
>>>>>>> supporting retro-logic, either implicitly or explicitly. The interaction
>>>>>>> between the FIDE laws and the retro-logics is more complex than between the
>>>>>>> Laws and the basic retro-conventions. Naturally,since the conventions are
>>>>>>> required input for the logics to function besides other concepts.
>>>>>>>
>>>>>>> Note: I hope to discuss some other day why I think that retro-logics
>>>>>>> and conventions should be separated. It doesn't matter for the current
>>>>>>> story that you subscribe to that view or not.
>>>>>>>
>>>>>>> One of the things the solution shows are the two different
>>>>>>> relationships between repetition and castling:
>>>>>>> 1. The dependency relationship dictated by the laws and conventions.
>>>>>>> Repetition may be dependent on castling right but not in reverse.
>>>>>>> 2. The (retro-)logical relationship between repetition and castling
>>>>>>> right. Here the choice of forward moves may decide on either a past with
>>>>>>> (loss of) castling rights or a drawing repetition. The context is usually
>>>>>>> one of retro-strategic logic.
>>>>>>>
>>>>>>> You can imagine that combining a retro-logic for trying to decide
>>>>>>> missing information from the past with a dynamic chess Law trying to look
>>>>>>> into an uncertain future from that past, at times produces quirky results.
>>>>>>> And so, besides solving the puzzle, I wil make a detour into Jumanji with
>>>>>>> the idea of "dynamic castling rights".
>>>>>>>
>>>>>>> Solution to the puzzle:
>>>>>>>
>>>>>>> Play forward: 5. Be2 Ng8 6. Bh5 Nf6 7. Be2 Ng8 The next white moves
>>>>>>> are not unique, the blacks are e.g. 8.Nc3 Rb8 9.Nb1 Ra8.
>>>>>>> By convention, the new position is not a draw unless a repetition
>>>>>>> draw can be proved. Such is not the case and we therefore have no draw
>>>>>>> here. From that information you can derive only 1 proof game without
>>>>>>> drawing repetitions: 1. Pe4 Na6 2. Bxa6 Nf6 3. Be2 Rg8 4. Bh5 Rh8. You can
>>>>>>> see that the eighth and ninth black move Rb8 and Ra8 demonstrate the type-1
>>>>>>> relationship (castling rights must have been lost there) and 7...Ng8 shows
>>>>>>> the type-2 relationship. Before that move, black could have had both
>>>>>>> castling rights, after it, one was lost. This shows the common paradoxical
>>>>>>> scene of losing castling right without ever moving King or Rook. Well, what
>>>>>>> you really lost, was just a "past". Happens all the time in retro-chess!
>>>>>>>
>>>>>>> Now to Jumanji. Suppose move 8 and 9 were not played in that game.
>>>>>>> Notice that before 7...Ng8 black had both future castling rights and after
>>>>>>> it he lost 1 - not to say which but that doesn't matter. Lets now play 8.
>>>>>>> Bh5 Nf6 again reaching the same position for the 3rd time. According to the
>>>>>>> "future castling law", the position is not the same as the previous times
>>>>>>> when there were still both future castling rights/occasions. In reality, by
>>>>>>> standard proof game assessment, the position is an unnegotiable draw at
>>>>>>> this point. Welcome to the Zoo!
>>>>>>>
>>>>>>> Explanation. The confusion comes from mixing retro-strategy and
>>>>>>> retro-analysis. Whenever the term "future" is used one reaches - almost
>>>>>>> unconciously - for what seems the appropriate retro-logic. Since there is
>>>>>>> no framework for handling repetitions in prA or retro-variants one resorts
>>>>>>> to the natural choice that is always available - retro strategy. But
>>>>>>> retro-strategy can eliminate roads into the past and thereby narrow down
>>>>>>> your (retro) options. Static definitions only refer to past proof games and
>>>>>>> need not take data from an ambiguous future. In this case, the proper
>>>>>>> retro-analyst arrives at the same conclusion Valery did in his example: a
>>>>>>> castling right was already lost by move 4 in all remaining proof games - as
>>>>>>> seen in retrospect after move 9.
>>>>>>>
>>>>>>> Well, I never said it is impossible to operate on dynamic rights,
>>>>>>> only that it requires an I.Q. of 280 plus a winning lottery ticket. Proves
>>>>>>> that you can manage the future.
>>>>>>>
>>>>>>> Guus Rol..
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Wed, May 14, 2014 at 6:47 PM, Guus Rol <grol33 at gmail.com> wrote:
>>>>>>>
>>>>>>>> Hi Joost and Valery,
>>>>>>>>
>>>>>>>> The way I see it there is a pre-issue to the discussion. The
>>>>>>>> FIDE-laws do not know of any automatic draws. Even in the simple case
>>>>>>>> without past or future "castling rights", FIDE must have considered that
>>>>>>>> "possible moves" would not change across 50M boundaries or repetitions
>>>>>>>> within repetition cycles. None of those enforce draws or close gateways
>>>>>>>> even when the options for the players do change - i.e. the right to claim.
>>>>>>>> Chess players though, are keenly aware of the risks in repeating
>>>>>>>> positions. You might say that when the board looks the same for the FIDE
>>>>>>>> laws, it looks quite different for the players. Example: Ke4-e5-e6-e5-e4
>>>>>>>> (mimicked by the opponent) not only switches back but also spoils the game
>>>>>>>> if Ke5 is now the only move to win it!
>>>>>>>>
>>>>>>>> Which means that we are left with the question on how to harmonize
>>>>>>>> the original FIDE concept with the addition of automatic draws in the world
>>>>>>>> of composition - still not considering castlings at all. I would suggest
>>>>>>>> that to "continue to ignore possible terminations" is most in line with the
>>>>>>>> FIDE concept. Yes, some repetitions do  change options for the players but
>>>>>>>> so does the original FIDE law in real chess games - see example above. In
>>>>>>>> fact, by down-playing the relevance of future developments in the
>>>>>>>> conversion from play to composition, we also take the first step towards a
>>>>>>>> static view on repetitions altogether. Do we need more then to justify the
>>>>>>>> application of static castling rights in repetitions as well?
>>>>>>>>
>>>>>>>> Guus Rol
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Wed, May 14, 2014 at 4:59 PM, Retros Probleemblad <
>>>>>>>> retro at probleemblad.nl> wrote:
>>>>>>>>
>>>>>>>>> On 05/14/2014 03:50 PM, Valery Liskovets wrote:
>>>>>>>>>
>>>>>>>>>> Joost,
>>>>>>>>>>
>>>>>>>>>> In my example, looking in the past, we have enough time for 4
>>>>>>>>>> repetitions.
>>>>>>>>>> But are they (more exactly, should the Rules be such that they
>>>>>>>>>> are) all
>>>>>>>>>> identical in both twins? I see no way to pose this question via a
>>>>>>>>>> problem
>>>>>>>>>> with an ordinary stipulation (unless I overlook anything). As far
>>>>>>>>>> as I know
>>>>>>>>>> Nikita Plaksin has never discussed this (modern, future-depending)
>>>>>>>>>> collision:
>>>>>>>>>> is castling _practically_ executable or not? And his
>>>>>>>>>> corresponding problems
>>>>>>>>>> didn't depend on such nuances of the 3-rep. rule. But maybe any
>>>>>>>>>> of them
>>>>>>>>>> can be reinterpreted in such framework?
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Valeri,
>>>>>>>>>
>>>>>>>>> Although the question is different, the underlying theme is the
>>>>>>>>> same: Is castling allowed if it can be proven that, by castling (or in your
>>>>>>>>> case: by making it possible to castle), more than 100 single moves without
>>>>>>>>> capture, pawn move or castling have occurred?
>>>>>>>>>
>>>>>>>>> Your example could be changed to something even more extreme by
>>>>>>>>> the way:
>>>>>>>>> Ke1 Rh1 Sa1 // Ke8 Ph2
>>>>>>>>> No captures/pawn moves for 46 full moves. How many times can the
>>>>>>>>> sequence Sb3 Ke7 Sa1 Ke8 be repeated until there is a correct claim for
>>>>>>>>> 3-fold repetition.
>>>>>>>>>
>>>>>>>>> Can white really lose his castling rights by moving a piece
>>>>>>>>> besides the rook and king?
>>>>>>>>>
>>>>>>>>> Joost
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Retros mailing list
>>>>>>>>> Retros at janko.at
>>>>>>>>> http://www.pairlist.net/mailman/listinfo/retros
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Retros mailing list
>>>>>>> Retros at janko.at
>>>>>>> http://www.pairlist.net/mailman/listinfo/retros
>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Retros mailing list
>>>>> Retros at janko.at
>>>>> http://www.pairlist.net/mailman/listinfo/retros
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> Retros mailing list
>>>> Retros at janko.at
>>>> http://www.pairlist.net/mailman/listinfo/retros
>>>>
>>>>
>>>
>>> _______________________________________________
>>> Retros mailing list
>>> Retros at janko.at
>>> http://www.pairlist.net/mailman/listinfo/retros
>>>
>>>
>>
>> _______________________________________________
>> 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/20140516/0b98561f/attachment-0001.html>


More information about the Retros mailing list