Website founded by
Milan Velimirović
in 2006

22:17 UTC
ISC 2022



Remember me

Forgot your
Click here!
to create your account if you don't already have one.

Rating lists


MatPlus.Net Forum General Why so few draw-problems?
You can only view this page!
Page: [Previous] [Next] 1 2
(1) Posted by Rosie Fay [Wednesday, Feb 10, 2021 11:56]

Why so few draw-problems?

Something I'd like a problem-composer's perspective on.

As my chess problem collection grows, and I find problems which stipulate a variety of chess phenomena as their goals, it is clear that one chess phenomenon is noticeably rare as chess problem goal: draw.

This is not to ignore draw-studies, which are a considerable percentage of all studies. But I wish here to concentrate on problems, with their specified length and tighter prohibition of duals.

Since about 1984, composers have composed problems with goals such as target-square, capture, check, pin, castling and unit-swap (Platzwechsel). So it is not as if checkmate as goal has become overwhelmingly popular. Perhaps composers think that solvers would prefer checkmating to drawing? Then I reply with John Nunn's words from Solving in Style, ch.5, p.74: "Somehow stalemate seems a less exciting theme than mate, perhaps because it is always more enjoyable to win than to draw. However, these subjective impressions have no effect on the score table, where a draw snatched from apparently certain defeat is just as beneficial as an unexpected win."

A mate-problem shows a battle between White, who is trying to win, and Black, who is trying to avoid losing. Composers show great skill in showing ingenious ways in which White can win against any defence. Any reason why they cannot do the same to show how the defending side can get a draw against any attack?

Perhaps there is this difficulty. The composer can enable the solver to prove that White wins by showing how White can checkmate against any defence. It is not always possible to prove so convincingly that the defending side can prevent the attacking side winning. They can deliver stalemate or reduce the attacking side to non-mating force, but it is not always possible to force this. It may in some cases be possible to force draw by repetition, but it is seldom possible to force this, except if the defending side can execute perpetual check.

Perhaps it's the nature of chess's tablebase in the borderlands between drawn positions and decisive ones. Perhaps positions where the turn player has several options only one of which wins are more amenable to problems than positions where the turn player has several options only one of which doesn't lose? By "amenable to problems" I mean that they are more plentiful, and are easier to find in sufficient numbers, and to connect to each other via best moves by the defending player. Typically, if I put a draw-study into Nalimov, it shows that White indeed has exactly one non-losing option, but after that, drawing options for Black and White are plentiful -- whatever subtlety the composer has put into their study, the tablebase doesn't show any more than the key move's uniqueness. It is the rarity with which the reverse occurs with drawn positions that makes Réti's famous KPkp study so marvellous even now, a century after Réti composed it.
(Read Only)pid=20608
(2) Posted by Joost de Heer [Thursday, Feb 11, 2021 11:40]

Winchloe has over 38000 compositions with aims '=', '==' or '!='.
(Read Only)pid=20610
(3) Posted by James Malcom [Thursday, Feb 11, 2021 14:09]

I think Rosie means studies, not fairy problems.
(Read Only)pid=20611
(4) Posted by Joost de Heer [Thursday, Feb 11, 2021 16:01]

No, she doesn't:


This is not to ignore draw-studies [.] But I wish here to concentrate on problems

Problem with non-study draws is that mate/zielfeld/castling/etcetc has a clear indicator that the goal is reached. But when is the goal 'draw' reached, if you don't mean stalemate in any form?
(Read Only)pid=20612
(5) Posted by Rosie Fay [Thursday, Feb 11, 2021 16:02]

I mean draw or stalemate problems (with specified length and tight standards against duals). I acknowledge that composers have composed many draw-studies. Also selfstalemates and series problems (seriesXstalemate where X might be null, help, self or reflex).
(Read Only)pid=20613
(6) Posted by Dmitri Turevski [Thursday, Feb 11, 2021 19:18]

I'm a bit confused too. What has to be reached against best defense in N moves if not stalemate or threefold repetition (insufficient material being just a special case of guaranteed threefold repetition)? A theoretically drawn position? But if the problem is sound, then the initial position is already theoretically drawn.
(Read Only)pid=20615
(7) Posted by Andrew Buchanan [Friday, Feb 12, 2021 04:15]

Almost all non-study “draws” are actually asking for stalemate. The ambiguity in the notation “=“ causes some confusion here. Stalemate is an exact concept, identical to mate except there is no check. Unlike in a study, a mate would not constitute an alternative solution or cook. Non-study stalemate problems are all classed as “fairy”.

In studies, 3-fold repetition is the default way a game is considered to end if no mate or pat can be forced. Outside studies it only occurs in retros, although.

50-move and dead position rules only apply by default in retros, although they can have applicability in forward compositions in dual elimination.
(Read Only)pid=20616
(8) Posted by Rosie Fay [Friday, Feb 12, 2021 08:32]

Ah, so perhaps that's it, then: the fact that a move achieves a draw is hard to prove, unless it delivers stalemate. (Or perpetual.) Thanks for your insights, folks.

Andrew Buchanan: Non-study stalemate problems are all classed as “fairy”.

In PDB, at any rate. More's the pity -- it isn't *necessary* to classify them so. As Mario Richter commented to PDB P1170859: 'it's a pity that there is no genre "=n", since imho labelling this problem as "Fairy" (at the moment the only possibility) is irritating and misleading: it suggests that the problem has some fairy condition or includes some fairy pieces.'
(Read Only)pid=20617
(9) Posted by Hauke Reddmann [Friday, Feb 12, 2021 09:57]

Well, it has a fairy stipulation then :-)
(Read Only)pid=20618
(10) Posted by Joost de Heer [Friday, Feb 12, 2021 10:20]

Stalemate compositions are in the fairy section of the FIDE album as well. The only magazine that I know of that has a separate stalemate section is Strategems, in all other magazines stalemates are also in the fairy section.

With mates, you only have to worry about one piece of one colour, with stalemate all pieces of one colour (and in double-stalemate all pieces of any colour).
(Read Only)pid=20619
(11) Posted by Andrew Buchanan [Friday, Feb 12, 2021 11:33]

The PDB genres are basically the album genres, except that a problem can belong to more than one. It’s also worth looking at the Codex. It’s sketchy in this area, and I heard that this was due to an inability to reach agreement at the time. Apparently Footnote 13 was very important as a political compromise:
13. In this context, the terms orthodox, heterodox, fairy and exo are used.
but I’ve never met anyone who can tell me what these terms mean. I interpret this that some people fought hard on a matter of principle, but actually what they fought over was meaningless - it was just important to them to win the argument.

To me, fairy can mean non-standard stipulation, as well as rules, board or pieces. But which stipulations? Helpstalemate seems a more plausible collaboration for two players than helpmate, but only the latter is orthodox. Direct stalemate is strictly less weird as a concept than selfmate but again only the latter is orthodox.

This is all so bogus. Some areas of any worthwhile design universe will be explored before others, but why partition the world rigidly based on a historical snapshot, when we know that the search for composition originality is bound to take everyone off into fresh design space? This partition wasn't itself a huge mistake: the problem was the rigidity, the lock-in.

The best that can perhaps be said is that the community of fairy folk is likely more diverse than if other terms were defined more generously.
(Read Only)pid=20620
(12) Posted by Peter Wong [Friday, Feb 12, 2021 12:48]

Typically, if I put a draw-study into Nalimov, it shows that White indeed has exactly one non-losing option, but after that, drawing options for Black and White are plentiful -- whatever subtlety the composer has put into their study, the tablebase doesn't show any more than the key move's uniqueness.

Just a minor correction here. While studies are perhaps more tolerant of duals compared with directmates, there's no difference between win and draw studies in this regard. In the main variation(s) of a draw study, every white move (not just the key) should be unique in forcing the draw, up to the part when the position becomes "clearly" (a bit subjective) drawn, e.g. material is dead even with no tactics available. Tablebases, like engines, are terrible at picking Black's thematic moves, so to test a study, you usually have to manually enter each of Black's moves in the main line and check if White's response is unique.
(Read Only)pid=20621
(13) Posted by Dmitri Turevski [Friday, Feb 12, 2021 14:39]

I think there is some kind of trade-off in dual tolerance between the problems and the studies.
While in studies the duals in non-thematic lines may be ignored, in problems the longer ways to checkmate are ignored.
Virtually any twomover is a valid though terribly cooked win study.
(Read Only)pid=20622
(14) Posted by Jakob Leck [Friday, Feb 12, 2021 16:03]

The title question seems to have been answered implicitly already.
Rosie, do you maybe underestimate the forces of convention, tradition and habit?

Personally, I would need a good reason to compose a helpstalemate instead of a helpmate when starting with a certain thematic idea. And, all things except the stipulation being equal, I would tend to prefer the h# because the h= is "fairy" (and thus in some sense less economical in terms of the means chosen to show the thematic idea). (Even though this is only a historically grown convention etc. etc.)
(Read Only)pid=20623
(15) Posted by Olaf Jenkner [Saturday, Feb 13, 2021 00:56]

I fully agree.
In many cases fairy elements are used to show themes that are impossible in orthodox way.
Everybody knows that the stipulation stalemate is a relief. So I see not only historical reasons for classify stalemates as a fairy stipulation.
It is easy to show the 100$ theme as a h=5. A Babson like this one does not exist in the helpmate genre:
(Read Only)pid=20624
(16) Posted by Andrew Buchanan [Saturday, Feb 13, 2021 05:33]

Indeed a composer will usually prefer h# to h= to display a random artistic idea (interference, cycles, allumwandlung etc). However some themes are better represented as stalemate, or can only be represented in that form. Pat is qualitatively different from mate, as a property of the whole board, not just the neighbourhood of bK. Judges and expert solvers are easily able to distinguish the motivation for the stipulation here. And why not have h=5 with double S excelsior? What's so terrible about that? We all know it's not the real deal.

There is no more justification for the ugly over-reaction that "*all* h= must be fairies" than there is so say that any problem with a severe design compromise is "fairy". I think the genre should be h, and keep the fairy tag as it's used in over-the-board play: for variant conditions, pieces & board. One can still have tourneys and magazine columns devoted solely to h#.

In my opinion, you are committing a common logical fallacy:

We all know that:
(1) A
(2) A implies B
allows us to infer:
(3) B

But are both (1) & (2) true? All too often the certainty with which (1) is held blinds one to the weakness of (2). Here A = "composers prefer to show their ideas in mate rather than pat form" while B = "stalemate should be fairy". There is zero connection between the two. "A implies B" is false here.
(Read Only)pid=20625
(17) Posted by Jakob Leck [Saturday, Feb 13, 2021 10:38]

Direct mates and studies are the prototypes of chess problems as they are directly derived from the game of chess. Thus they were and still are naturally orthodox.
Other genres like helpmates or selfmates used to be considered as "fairy" and only through their growing popularity (and their very small deviation from the orthodox stipulation) have they come to be regarded as genres of their own, "orthodox" helpmates and selfmates.
This is what I mean by the historical reasons for the "fairy" classifications.
And of course we can imagine the course of history having been different - which seems to render the classification somewhat random - but it hasn't been!

In this sense "all h= are fairies" is not an over-reaction, it's just a fact.

And while there is, of course, nothing at all wrong with choosing to compose (h)= instead of (h)#, there are good reasons for composers' reluctance to deviate from the "orthodox" ways: economy of means (in the given context outlined above), comparability of problems in tournaments, relying on large experience and sharpened skills of both composers and solvers (in an environment that has been studied more in depth), possibly a wider audience etc.
(Read Only)pid=20626
(18) Posted by Hauke Reddmann [Saturday, Feb 13, 2021 11:18]

Another aspect: How do you compose a problem anyway?

I bet you first have some idea. Now, ideas do not yet
come by drone from :-), they rather come
from looking at other problems (unless you have a
very VERY original brain). And bang, Polya urn.

(Read Only)pid=20627
(19) Posted by Andrew Buchanan [Saturday, Feb 13, 2021 11:20]

@Jacob: when I saw you had replied, I thought great! I really want to read an argument that convinces me. I don’t claim to be right at all. But I don’t see any idea in what you wrote except a defence of and dependence on the status quo. It’s all circular. And hence subjective - what is “natural” to you is “contrived and ugly” to me. My whole point is that the status quo must change, because that’s the nature of the quest for originality. The Codex should define the stakeholders whose interests are equally protected: largely composers and solvers but also judges and authors, from the past, present and future. An unmovable notion of “good” and “bad” locks in a prejudice against the future. As footnote 13 proves there is no real meaning to the status quo: it’s just loved by some and hated by others because it’s the status quo. Please try again - I don’t want to be right thanks.
(Read Only)pid=20628
(20) Posted by Rosie Fay [Saturday, Feb 13, 2021 11:24]

Thank you, Andrew, for your information about FIDE Album categories. The consequences for me as PDB-user is, sadly, that I find it hard to find problems with stipulations other than #, h#, s# and r#, because I have to use a search which allows problems with genre Fairies, which lets in many with fairy conditions or fairy units. I generally end up explicitly excluding stipulations containing certain letters of the alphabet (which is not perfectly effective), and I always worry about what extra load I'm putting on PDB's server by making my search criterion more complicated.

To PDB's maintainers' credit they have applied keywords such as 'Series mover' and 'Non-standard material'. Now if only there were a keyword 'Non-standard chess' ... Yes, unfortunately this applies to many tens of thousands of PDB entries, and I won't underestimate the scale of the work that would be involved. Just pointing out the trouble caused by the consequences of the present classification.

Jakob Leck, even convention, tradition and habit are subject to challenge! And as we see, there are many ways in which a problem can depart from the orthodox, and different composers have chosen different combinations of the orthodox and the fairy. Which is all to the good -- it's by inventing new fairy concepts and trying to compose problems exploiting them that we see to what extent the composers can show effects that justify the fairy idea.

Another possible reason to compose a problem with stalemate goal: It's one peculiar consequence of the geometry of the chessboard and the power of the chess pieces that although selfmate and seriesselfmate can work where White has KQ, and also where Black has K + 1, at least 5 units are needed: they can't work with just K + 1 v K + 1. But selfstalemate and seriesselfstalemate can. (Pity P1091590 and P1091589 are unsound, though.)
(Read Only)pid=20629

Page: [Previous] [Next] 1 2

MatPlus.Net Forum General Why so few draw-problems?