MatPlus.Net

 Website founded by
Milan Velimirović
in 2006

14:20 UTC
ISC 2024
 
  Forum*
 
 
 
 

Username:

Password:

Remember me

 
Forgot your
password?
Click here!
SIGN IN
to create your account if you don't already have one.
CHESS
SOLVING

Tournaments
Rating lists
1-Jan-2024

B P C F





 
 
MatPlus.Net Forum Competitions Difficulties with points estimation in solving competitions
 
You can only view this page!
Page: [Previous] [Next] 1 2
(21) Posted by Georgy Evseev [Monday, May 25, 2009 14:00]

@ Jacques & Miodrag

Your are right - this is how such case should be judged. But will you still be sure of your decision, if director's solution looked like

1...Sa3/Sc3 2. A
1...Sd2 2. B

(this still has exactly the same meaning).


About your discussion about "versions" given by solver. There is nothing in current rules which allows solver to record several versions of correct move. Traditionally, such records are ignored altogether (as special case of ambiguity which should not be allowed). Though I would agree that this case should be mentioned in rules explicitly.

But sometimes the solver writes down several threats... and sometimes he is even right doing this... And here we again enter a maze of vague, informal and non-obvious cases...
 
   
(Read Only)pid=3716
(22) Posted by Jacques Rotenberg [Monday, May 25, 2009 14:29]; edited by Jacques Rotenberg [09-05-25]

"...1...Sa3/Sc3 2. A
1...Sd2 2. B

(this still has exactly the same meaning)."

What is the question?

This allows 2 points
 
   
(Read Only)pid=3717
(23) Posted by Georgy Evseev [Thursday, May 28, 2009 10:18]

Here is another, more difficult case for decision (I still hope to produce a case which will cause different opinions).

The example is the same as before - in a threemover all moves of black knight b1 are defenses and solvers have given the key correctly.

Director's part of relevant variations is also same as before:

1... S~ 2. A (1 point)
1... Sd2 2. B (1 point)

Solvers has written:

a)
1... S~ 2. A
1... Sc3 2. B

b)
1... Sa3/Sd2 2. A
1... Sc3 2. B

c)
1... S~ 2.B

Please write your opinions about correct judgment (my opinion has _changed_ during thinking about these situations).
 
   
(Read Only)pid=3744
(24) Posted by Miodrag Mladenović [Thursday, May 28, 2009 12:58]

@Georgy(28)

When I applied the current rules aginst all three scenarios I am getting 1 point (a, b & c). This is under an assumption that there is only one black knight on the board. In all three scenarios there is only one good variation and the other one is wrong. The tricky one is the third one. Since S~ does include Sd2 I think that per current rules it's one point. And of course it's obvious that solver did not see solution properly.
 
 
(Read Only)pid=3745
(25) Posted by Hauke Reddmann [Thursday, May 28, 2009 16:21]

Dear Georgy -
shouldn't the last example read d) 1...S~ 2.A?

:-)))

Hauke
 
   
(Read Only)pid=3746
(26) Posted by Georgy Evseev [Friday, May 29, 2009 12:52]

@Hauke

No, it was not a mistake. That case was discussed before.

@Miodrag(24)

Please, comment on the following logic. (I think it is almost correct, but I know at least two different possible objections and also some unexpected conclusions may follow.)

-----
The director's solution in practice unites three different variations:

1... Sa3 2. A
1... Sc3 2. A
1... Sd2 2. B

The first two variations are additionally united by the director because white play is the same and score 1 point together.
In all three cases the solver wanted to cover all three variations.

In cases a) and b) really only variation 1... Sa3 is shown correctly. So, only half of the set of two variations is correct, so director should penalize solver for error - and it is logical to give him only 0,5 points.

In case c) again only one variation is correct and it is worth 1 point, so 1 point is what the solver receives.

-----

Here is the pertinent part of rules:

"8. The solutions are to be written in the following way:
a) in direct mate problems: all moves in all variations of full length which defend the threat, including the threat if it is of full length, except the last move of Black and the mating move (i.e. in twomovers only the key; in threemovers all three move variations, including the full length threat, up to the second white move, etc.);
b) in selfmate problems: all moves except the mating move in all variations of full length which defend the threat, including the threat if it is of full length.

If a threat doesn't exist or a threat of full length is not written all variations of full length are to be given.

c) in helpmate problems: all moves;
d) in endgames: all moves up to an obvious win or draw."
 
   
(Read Only)pid=3748
(27) Posted by Georgy Evseev [Monday, Jun 1, 2009 08:20]

Here is another case which happens very often, but still has no official ruling.

Director's solution:

1...a,b 2. A... (1 point)

Solver's record:

1...a 2. A...


Current status.

There are two different opinions, I can call "composer's" and "solver's".

"Composer's" opinion: the solver has found the necessary _variation_ and should receive full score.

"Solver's" opinion: the solver has not found all the _branches of solution_ and should be penalized.

The current rules (see post 26) does not strongly support any of this opinions and allows to argue ad infinitum. Formally, both these opinions do not contradict current rules.

Also if a decision to penalize solver is accepted there is a question: how it should be done. There is a lot of possibilities of different solver's mistakes. He can add wrong defenses, the number of correct defenses can be bigger or smaller (and different number of defenses can be shown correctly) and so on. Currently, possible judges' decisions are not unified.


My opinion.

I think we should add two formal definitions to our rules, concerning threemovers, moremovers and selfmates. (I hope someone will be able to find a good wording.)

"Variation" should be defined as the sequence of moves starting from move of white and allowing to reach the goal in required number of moves. This definition can be considered recursive for branching variations.

"Defensive group" should be defined as a set of black moves allowing the same variation to occur.

The scoring system should include a small penalty for solver's mistake. That is, I would use "solver's opinion", in the form as near to "composer's" as possible. Something like this.

"If solver has found the variation and has correctly shown the defensive group he receives full score for this variation. If he has correctly shown at least one black defense and has also missed at least one defense from this defensive group, then 0.5 points is subtracted from full points for this variation."

If these suggestions were in place then solver in the example above would have received 0.5 points (0 points if the variation was worth 0.5 points, 1.5 points if the variation was worth 2 points).
 
   
(Read Only)pid=3759
(28) Posted by Miodrag Mladenović [Monday, Jun 1, 2009 09:01]

@Georgy(27)

I like your proposal but still there will be issues. The problem is if #3 does contain more then 5 variations (including threat). Then, there could be variation like this one that originally is given 0.5 points (not 1 or 2). Now if you do subtract 0.5 points for omitting of one black defense solver will end up with 0 points for variation what is also wrong.

Imaging situation where there are two #3 where there is a same white variation with multiple black defenses. In the first threemover because there are fewer variations solver will get 0.5 points instead of 1.0 (assumption is that maximum point for this variation is 1.0 point).
In the second #3 there could be more variations so the full points for similar variation are 0.5. Now if solver missed one black defense he will get 0.0 points even if he saw variation with some other defenses.

This will mean that second solver will be penalized more because he is not going to get any points for something that he solved. Perhaps better rule will be to divide designated points for variation with a number of black defenses and deduct points based on this fact. With today’s technology it’s easy to calculate points like 0.33 etc.
 
   
(Read Only)pid=3760
(29) Posted by Georgy Evseev [Monday, Jun 1, 2009 10:51]

@Miodrag(28)

Unfortunately, there always will be issues because different solvers/composers/judges have different opinions and under current rules all these opinions are correct. I understand your concern (and I even mentioned this situation in parenthesis), but I think this fixed penalty of 0.5 points is better decision than direct calculating of fractions.

First, my suggestion takes into account point of view marked above as "composers". Sometimes it is really difficult to find a variation itself and giving a minor fraction does not seems correct. (See the example in the end of the post.)

Second, the use of minor fractions put into doubt one of crucial current principles of judgment, which says: points first, time last. If we beginning to use minor fractions than this becomes not well justified. Why 0.0001 of point should count more than 30 min less time? (This is, of course, exaggeration, but still valid argument. See also post 6.)

Third, a fixed penalty of 0.5 point for this kind of error is suggested. It is considered by me a minimal penalty possible (again see post 6).

------
Example.

I planned to show this example later, because it deals with threat, but we are going to move to discussing threats anyway.

This example is from real competition. Names of participants and judge are currently withheld, but the decision could have influenced who will be the overall winner.

Last round. The problem in question is S#6. Black has six different moves after the key. There are two exact moves with full variations (worth 2 points each) and a threat (works for all four remaining moves, and worth 1 point.)

The solver in question has written:

1... a 2. A... (and so on, all threat moves correctly).

Move "a" is really one of four "random" moves which do not defeat the threat. Variations with exact defenses were written correctly. The competing solver has found the key and both other variations but missed the threat altogether.

The judge decided to give 0.5 points to the solver in question for this variation. Was he correct?
 
   
(Read Only)pid=3763
(30) Posted by Neal Turner [Monday, Jun 1, 2009 14:22]

No I don't think he was correct.
The solver put down a move followed by the correct variation.
Asking oneself 'Has he solved the problem?', the answer must be 'Yes' and he should get the full points.
As was mentioned earlier, there is currently nothing in the rules that allow for penalty points in the case of minor omissions or mistakes - so if the judge wanted to penalise the solver he should have withheld the full point, which would have been ridiculous.
 
   
(Read Only)pid=3764
(31) Posted by Georgy Evseev [Monday, Jun 1, 2009 15:03]; edited by Georgy Evseev [09-06-01]

@Neal

You have just repeated what I've called "composer's opinion" in post 27.
As the solver has not shown any threat, we can (should?) interpret the judges' solution as

1...a,b,c,d 2. A... where a,b,c,d - neutral black moves allowing the threat to work.

And the solver's record was

1...a 2. A... (compare with post 27)

Representing "solver's opinion": "Asking oneself 'Has he solved the problem?', the answer must be"... "No", as the solver has not provided correct continuations for several black moves available after the key.

But I'll agree with you that your opinion is correct because it also cannot be rejected. And this is the main difficulty as I've stated from the beginning, as different decisions are equally available under current rules.
 
   
(Read Only)pid=3765
(32) Posted by Georgy Evseev [Tuesday, Jun 2, 2009 09:01]

@Neal

After some thinking I have decided that maybe your suggestion from post 30 is best. Not because it is more correct than other ideas, but because it allows to build formal, reasonably simple and logical method for evaluation. I think I now have the reasonable judgment suggestion (for #3, #n and s# sections), but still want to show all my remaining "examples" first.

Here is another example (from two remaining)

---------

The situation is typical for selfmates.

After the key there is a zugzwang for black. They have exactly three moves and each leads to a separate variation, like this...

1...a 2. A... (let it be 2 points for this variation)
1...b 2. B...
1...c 2. C...

Specifically, none from these variations works as a threat.

There are two solvers who have the correct key and the following records:

A)

1...~ (or threat) 2. A...

B)

1...~ (or threat) 2. A...
1...b 2. B...
1...c 2. C...

How these records should be evaluated? Will this evaluation change if a virtual threat 2. D... exists but defeated by all black moves?
 
   
(Read Only)pid=3770
(33) Posted by Georgy Evseev [Wednesday, Jun 3, 2009 09:33]

This is the last example for this part of discussion, again from real competition. The key was shown correctly. The relevant variations in Director's solution were:

1...~ 2. A... (1 point)
1...x 2. B... (1 point)

The solver has written

1...~ 2. B...
1...a,b,c (second threat) 2. A...

The moves a,b,с were considered by solver as defeating the "first threat" (which really was not a threat). In reality they are neutral moves allowing "real threat" to happen. Defense "x" was not specifically mentioned by solver.

What should be a correct decision in this case? I can in accordance with current rules logically motivate any result from 0 to 1,5 points.
 
   
(Read Only)pid=3774
(34) Posted by Jacques Rotenberg [Wednesday, Jun 3, 2009 21:25]; edited by Jacques Rotenberg [09-06-03]

...
1...~ 2. A... (1 point)
1...x 2. B... (1 point)

The solver has written

1...~ 2. B...
1...a,b,c (second threat) 2. A...

very simple :

1...~ 2. B... --> 0
1...a,b,c (second threat) 2. A... --> 1

NO NEGATIVE POINTS !!

To add negative points seems to me to be very .... negative !
Why ?
Because, as you explain in details there are many cases that are not so easy, so it seems not convenient to make the task of the judge even more complicated.
 
   
(Read Only)pid=3780
(35) Posted by Georgy Evseev [Thursday, Jun 4, 2009 07:35]

@Jacques

Currently, I'm thinking that your suggestion is correct. But I'll not agree that this is "very simple".

Before this discussion I would have called correct another decision, and in practice the judge of competition had the third opinion. That is why I think some kind of standard is needed.
 
 
(Read Only)pid=3781

No more posts
Page: [Previous] [Next] 1 2

MatPlus.Net Forum Competitions Difficulties with points estimation in solving competitions