SearchSearch   MemberlistMemberlist   UsergroupsUsergroups  ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Adjustement needed for square rule scouts
Goto page 1, 2  Next
 
Post new topic   Reply to topic    Gravon - Das Spielerparadies Forum Index -> General
View previous topic :: View next topic  
Author Message
ruben87
Gravon Administrator


Joined: 16 Feb 2006
Posts: 1220
Location: The Netherlands

PostPosted: 10.08.2017 13:28    Post subject: Adjustement needed for square rule scouts Reply with quote

I always thought Gravon had all square rules perfectly implemented. But today I experienced in a test with Losermaker that there is one flaw in the square rule for scouts. I think Dieter already heared about this from Gary from the other site. The following is not working properly

0) Start at A1 (not from line A to this square)
1) move from A1 to A5
2) move from A5 to A3
3) move from A3 to A7
4) move from A7 to A2 (should be not allowed to go to a2/a3/a4)

at number 4 it should not work to go to a2,a3 and a4 but it has to be allowed to go to a5 and a6. Gravon isnt allowing to go to a5 wrongly. So a little bug here is still not fixed.

Can some admin fix this please?
Back to top
View user's profile Send private message
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 02.01.2021 11:14    Post subject: Re: Adjustement needed for square rule scouts Reply with quote

ruben87 wrote:
I always thought Gravon had all square rules perfectly implemented. But today I experienced in a test with Losermaker that there is one flaw in the square rule for scouts. I think Dieter already heared about this from Gary from the other site. The following is not working properly

0) Start at A1 (not from line A to this square)
1) move from A1 to A5
2) move from A5 to A3
3) move from A3 to A7
4) move from A7 to A2 (should be not allowed to go to a2/a3/a4)

at number 4 it should not work to go to a2,a3 and a4 but it has to be allowed to go to a5 and a6. Gravon isnt allowing to go to a5 wrongly. So a little bug here is still not fixed.

Can some admin fix this please?


I don't think anyone posted this, but this bug has not been fixed yet. It should be legal to play 4) a7-a5. The easiest way to implement the two squares rule for scouts is to track the square borders that have been crossed in the last 3 moves. The borders between a3/a4 and a4/a5 have been crossed 3 times in the first 3 moves, so cannot be crossed again on the 4th move. But from a7 to a5 there is no such objection.
Back to top
View user's profile Send private message
stratego
Chief-Admin


Joined: 20 May 2003
Posts: 1123
Location: Germany

PostPosted: 03.01.2021 20:39    Post subject: Reply with quote

Hi,

i think i talked with spion about the rule a long time ago.
Its added to the the "to do list"

Beside i can offer a big approvement to StraDoS. Spion is working at some
very nice features for us all. This will be awesome and i dont think any other
place has something else.

greets
Stratego
Back to top
View user's profile Send private message Visit poster's website
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 06.01.2021 19:57    Post subject: Reply with quote

Good to hear this scout bug is on the repair list, and also that more StraDoS features might be coming.

Apropos rule bugs: there is also a corner case of the More Squares Rule not being correctly handled here. The documentation here states:

http://www.gravon.de/gravon/stratego/msr.jsp?language=english

in the last bullet:

Quote:
ALL situations previously on the board do count, including situations long before the first threatening (until a battle occurs, see above)!


I think the correct interpretation should be that only positions that occurred previously during the current chasing (=threat+evasion) sequence cannot be repeated. After every non-evading move (and not just after a capture move) the history list should be erased again.

This interpretation of the More Squares Rule is the one used by e.g. Vincent de Boer in his thesis on Computer Stratego (see p. 113 of his thesis here: http://www.kbs.twi.tudelft.nl/docs/MSc/2007/deBoer/thesis.pdf)

There is also an old riddle in the Marschall Plan "Problematisch" section (nr. 2001/2) that is not correctly handled currently on this site.

I'm curious to hear your opinion on this matter.
Back to top
View user's profile Send private message
stratego
Chief-Admin


Joined: 20 May 2003
Posts: 1123
Location: Germany

PostPosted: 06.01.2021 21:58    Post subject: Reply with quote

Hi,

maybe a german/english translation prob?

i dont know what problem that was.
but we must have in mind, that a chaser may not play a chasing move which would lead to a position on the board which has already taken place.

11. Repetition of Threatening Moves: More-Squares Rule

11.1 Itis not allowed to continuously chase one or more pieces of the opponent endlessly. The continuous chaser may not play a chasing move which would lead to a position on the board which has already taken place.

11.2 Exception: chasing moves back to the square where the chasing piece came from in the directly preceding turn are always allowed as long as this does not violate the Two-Squares Rule (Three-Moves-on-Two-Squares).

11.3 Definitions: continuous chase: the same player is non-stop threatening oneor more pieces of his opponent that is/are evading the threatening moves. chasing move: a move in a continuous chase that threatens an opponent‟s piece that was evading during the continuous chase. Hereby: a/to move: a/to move plus attacking or a/to move to an empty square. to threaten: to move a piece next (before, behind or besides) a piece of the opponent. to evade: to move a piece away in the direct following move after it has been threatened.
Back to top
View user's profile Send private message Visit poster's website
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 07.01.2021 16:49    Post subject: Reply with quote



OK, concrete example: in the above diagram it's Red to move.

1. a2-a3 (chase) b3-b2 (evade)
2. b1-c1 (chase) c2-c3 (evade)
3. a3-a2 (chase) b2-b3 (evade)
4. c1-b1 c3-c2 <--------- chasing sequence stops here
5. a2-a3 (chase) b3-b2 (evade)
and now the site does not allow red to play 6. b1-c1 again because the position after Red's 2nd move would be repeated.

However, the chasing sequence had been stopped on the 4th move already. So it appears that Red cannot defend this position. Nevertheless, there is no endless chasing as each 4th move will be a non-threatening move.

Can you indicate whether or not this site is correct about refusing 6. b1-c1 in this sequence?
Back to top
View user's profile Send private message
stratego
Chief-Admin


Joined: 20 May 2003
Posts: 1123
Location: Germany

PostPosted: 07.01.2021 21:44    Post subject: Reply with quote

Hi,

yes you are right, move 4 reset the sequence.

the chasing move must be continuously.

greets
Dieter
Back to top
View user's profile Send private message Visit poster's website
spion
Gravon Administrator


Joined: 27 Feb 2002
Posts: 750
Location: Koblenz

PostPosted: 10.01.2021 20:51    Post subject: Reply with quote

Hi,

i agree with you that the Gravon server is not following the rules 1:1 in both situations, in the two squares rule situation with scout as well as in your more squares rule with four scouts.

Thing is, it's not a bug in Gravon's server. Back in 2003-4, the time i developed the code for both rules, i had a different interpretation of the rules. Either the rules have been rewritten since then or i interpreted them in a wrong way, i don't know.

I had a discusstion with Dieter last week about the two squares rule problem. Back in 2003, i interpreted the rules as this way:

If the scout moves back and forth in the same line or column on the board in three moves and tries to do so in a fourth move, he must not move to a field he already "touched" or "crossed" in the three preceding moves. In this interpretation, the scout would not be allowed to move to A5, because he touched or crossed this field in all three preceding moves.

Last week, Dieter and me read the rules again:

Current ISF rules, 10.2:
Quote:
When a scout is involved in the Two-Squares
Rule, a scout is considered to start on the
starting position of his move plus all the
squares he steps over, and he ends on the
final position of his move plus all the squares
he steps over.


Let me try to re-write this rule in a mathematical form:

Definition:
The "source field" is defined as the field the scout was located before it's move.
The "target field" is defined as the field the scout is located after it's move.
A "crossed field" is defined as every field, that is located between the "source field" and the "target field" not including the "source field" and the "target field" itself.

Example:
Scout moves from A1 to A6. Then A1 is the "source field" and A6 is the "target field" and A2,A3,A4 and A5 are "crossed fields".

Definition, continued:
If the scout moves from "source field" F1 to "target field" F2, fields F1 and all "crossed fields" are defined as "starting fields". Field F2 and all "crossed fields" are defined as "ending fields".

In the example:
A1 to A5 are "starting fields". A2 to A6 are "ending fields".

A1: S
A2: S/E
A3: S/E
A4: S/E
A5: S/E
A6: E

Rule:
A scout must not move from field F1 to field F2 if a pair of fields S1+S2 exists that can be found as "starting field" S1+"ending field" S2 in the move from F1 to F2 and the move M-2 and that can be found as "starting field" S2 + "ending field" S1 in moves M-1 and M-3, where M-1 is the last valid move of this player, M2 is the 2nd last valid move of this player and M-3 is the third last move of this player and the same scout was moved in all these three moves.

This rule allows the scout to move to A5 and so either my original interpretationion of the rule was wrong or the rule was changed since 2004.

Either way, it would be much easier if the rule just was:

"The scout (as any other piece) must not step over the same border between two fields F1 and F2 during it's move for more than 3 consecutive moves."


Now about the more squares rule: I'm currently discussing that with Dieter too but here again i think i interpreted the rule in the wrong way or it changed...

At the moment, Gravon forbids any threatening move if this move would lead to a situation that already took place. Thereby, i ignored the definition of "chasing move" that says: "a move in a continuous chase that threatens an opponent‟s piece that was evading during the continuous chase".

So according to the rules, not only the situation must have taken place but the threatening move must threaten a piece that evaded in the same continuous chase before. In your situation, a piece is threatened that evaded in the previous continuous chase but this very piece did not evade in the current (second) continuous chase yet. So the move should be allowed.

So effectivly, the server must mark all pieces during a continuous chase that were threatened and evaded. A move is then forbidden if and only if such a marked piece is threatened and the very same situation already had taken place anytime throughout the game. The marking is reset on any non-threatening or non-evading move.

Best regards,
Thorsten
Back to top
View user's profile Send private message Send e-mail Visit poster's website
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 11.01.2021 11:47    Post subject: Reply with quote

spion wrote:

Rule:
A scout must not move from field F1 to field F2 if a pair of fields S1+S2 exists that can be found as "starting field" S1+"ending field" S2 in the move from F1 to F2 and the move M-2 and that can be found as "starting field" S2 + "ending field" S1 in moves M-1 and M-3, where M-1 is the last valid move of this player, M2 is the 2nd last valid move of this player and M-3 is the third last move of this player and the same scout was moved in all these three moves.

This rule allows the scout to move to A5 and so either my original interpretationion of the rule was wrong or the rule was changed since 2004.


I agree 100% that this is how it can be determined. Rule 10.2 just expands the definition of starting and ending squares to that you can find a sequence S-E E-S S-E E-S that is not allowed.

Quote:

Either way, it would be much easier if the rule just was:

"The scout (as any other piece) must not step over the same border between two fields F1 and F2 during it's move for more than 3 consecutive moves."

Again, I agree 100% that this would be a much simpler formulation of ISF rule 10, that would merge 10.1 and 10.2 into a single unambiguous article.

BTW, Vincent de Boer also mentions the "no more than 3 consecutive crossings of a single edge square by the same piece" formulation (p.111 of his thesis, that I linked to in an earlier post in this thread). Although at the time of writing his thesis, it was still 5 instead of 3 consecutive moves.

Vincent also mentions a cheap way of computing this. If the consecutive moves with the same piece take place in the same column, you can apply the following procedure. Let's apply it to Ruben's sequence a1-a5 a5-a3 a3-a7 a7-a5.

1. For each move, sort the pair of y-coordinates from low to high. You get 4 ordered pairs: (1, 5), (3, 5), (3, 7), (5, 7).
2. Now take the minumum of the 4 highest y-coordinates within each pair: min(5, 5, 7, 7) = 5
3. Also take the maximum of the 4 lowest y-coordinates within each pair: max(1, 3, 3, 5) = 5.
4. If the result from step 2 <= the result of step 3, the move is allowed.

Here 5 <= 5 and so the move is allowed. If the 4-th move had been a7-a4 instead, you would get the same result 5 in step 2 but max(1, 3, 3, 4) = 4 in step 3. So the move would not have been allowed since 5 > 4. And indeed the border between a4 and a5 would have been crossed by the same piece on 4 consecutive moves.

For 4 consecutive moves with the same piece on the same row, simply apply the above procedure on the x-coordinates instead.
Back to top
View user's profile Send private message
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 11.01.2021 13:21    Post subject: Reply with quote

spion wrote:

Now about the more squares rule: I'm currently discussing that with Dieter too but here again i think i interpreted the rule in the wrong way or it changed...

At the moment, Gravon forbids any threatening move if this move would lead to a situation that already took place. Thereby, i ignored the definition of "chasing move" that says: "a move in a continuous chase that threatens an opponent‟s piece that was evading during the continuous chase".

So according to the rules, not only the situation must have taken place but the threatening move must threaten a piece that evaded in the same continuous chase before. In your situation, a piece is threatened that evaded in the previous continuous chase but this very piece did not evade in the current (second) continuous chase yet. So the move should be allowed.

So effectivly, the server must mark all pieces during a continuous chase that were threatened and evaded. A move is then forbidden if and only if such a marked piece is threatened and the very same situation already had taken place anytime throughout the game. The marking is reset on any non-threatening or non-evading move.

Best regards,
Thorsten


I don't think I agree with the above explanation. Let's slightly expand my original example (same diagram, but repeating the first 4 moves a 2nd time):

Code:
1. a2-a3 (threatens the B-scout)  b3-b2 (evades the B-scout)
2. b1-c1 (threatens the C-scout)  c2-c3 (evades the C-scout)
3. a3-a2 (threatens the B-scout)  b2-b3 (evades the B-scout)
4. c1-b1 (continuous chase stops) c3-c2
5. a2-a3 (threatens the B-scout)  b3-b2 (evades the B-scout)
6. b1-c1 (threatens the C-scout)  c2-c3 (evades the C-scout)
7. a3-a2 (threatens the B-scout)  b2-b3 (evades the B-scout)
8. c1-b1 (continuous chase stops) c3-c2


So if I'm understanding you correctly, the 5th move (that is currently not allowed on Gravon) should be allowed, as well as the 6th, but the 7th move should not? And your reasoning would be that since in Red's 7th move, he threatens the B-scout that had previously evaded on the 5th move, resulting in a position that already occurred after Red's 3rd move, correct?

I think that's the most literal interpretation of ISF Rule 11.1, but I think a case could be made that it is supposed to be interpreted as follows:

Quote:
11.1
It is not allowed to continuously chase one or more pieces of the opponent endlessly. The continuous chaser may not play a chasing move which would lead to a position on the board which has already taken place during the current continuous chase.


So Red's 4th move resets the entire history of positions that cannot be repeated during a continous chase. In other words, the block of the first 4 moves can be repeated endlessly, all 8 moves shown should be legal, and the diagram is a draw.

The "during the current continuous chase" interpretation is how Vincent de Boer describes the More Squares Rule (same text as above, from the current official ISF page). He resets the position history for a chasing player after every non-chasing move (see p.113 of the aforementioned thesis).

I wonder how the diagram would be judged by a referee in an official live tournament?

kind regards,

Rein
Back to top
View user's profile Send private message
spion
Gravon Administrator


Joined: 27 Feb 2002
Posts: 750
Location: Koblenz

PostPosted: 11.01.2021 21:49    Post subject: Reply with quote

Hi,

i don't agree with the addition "during the current continuous chase". This part of the sentence simply isn't there in the rules. The sentence in the rules is imho totally clear: The situation must never have taken place. No matter at what point in time.

And i think it's good this way. I don't think that there is a need in allowing the same situation to happen twice. Do you?

Best regards,
Thorsten
Back to top
View user's profile Send private message Send e-mail Visit poster's website
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 12.01.2021 09:57    Post subject: Reply with quote

spion wrote:
Hi,

i don't agree with the addition "during the current continuous chase". This part of the sentence simply isn't there in the rules. The sentence in the rules is imho totally clear: The situation must never have taken place. No matter at what point in time.


Agreed that the phrase "during the continuous chase" is absent in the ISF rulebook. But the explicit phrase "during the entire history of the game" is also missing in Rule 11.1. In my opinion, that leaves some ambiguity to how it should be interpreted.

Quote:

And i think it's good this way. I don't think that there is a need in allowing the same situation to happen twice. Do you?


For online Stratego, the Gravon implementation is fine. The server can perfectly track the entire game history since the last capture.

But in my (limited) experience in live play, it's extremely hard for human players and referees to track history between multiple chasing sequences. From what I have seen, referees will judge that any non-chasing move will reset the history for the purposes of the More Squares Rule. This make it much easier to reason about practical board situations.

And Vincent de Boer also uses this interpretation of the More Squares Rule in his thesis and program. He could easily have programmed it like it has been done here. But apparently it was his experience as well how this is done in live play. I would find it surprising that a former world champion would have incorrectly applied the official rules (I know, it's an argument to authority, but still).

I'm curious how Dieter would judge such a position as a referee under ISF rules in live tournament play?
Back to top
View user's profile Send private message
spion
Gravon Administrator


Joined: 27 Feb 2002
Posts: 750
Location: Koblenz

PostPosted: 13.01.2021 23:46    Post subject: Reply with quote

Hi,

in the end, it's more easy for me if only situations of the current chase are taken into account.

I would propose the following algorithm, perhaps you could check it?

Algorithm:

If a player A makes a threatening move, mark him as "chaser". If, at any time, player A does not make a threatening move or player B does not make a move that evades the preceding threat, reset A's mark.

As long as player A is marked, do:

Whenever player B moves a piece, mark this piece as "evaded". If player A tries to do a threatening move AND the threated piece is marked as "evaded" AND the situation on the board is identical to a situation that took place before while player A was marked as "chaser" AND this move is not the exact opposite of A's preceeding move, the move is rejected.


I think this would follow the rules exactly.

The following would not be forbidden by this algorithm and IMHO also not by the rules:

I move my marshall so that it threats your general. You move your marshal so that it threats my general. I move my general away. You move your general away. And i start again by threatening your general. This would be allowed endlessly because none of us is continuously chasing. We always interrupt that because we both threat and both evade. Correct?

Best regards,
Thorsten
Back to top
View user's profile Send private message Send e-mail Visit poster's website
templaterex
User


Joined: 15 Dec 2017
Posts: 16
Location: Netherlands

PostPosted: 14.01.2021 21:59    Post subject: Reply with quote

spion wrote:

Algorithm:

If a player A makes a threatening move, mark him as "chaser". If, at any time, player A does not make a threatening move or player B does not make a move that evades the preceding threat, reset A's mark.

As long as player A is marked, do:

Whenever player B moves a piece, mark this piece as "evaded". If player A tries to do a threatening move AND the threated piece is marked as "evaded" AND the situation on the board is identical to a situation that took place before while player A was marked as "chaser" AND this move is not the exact opposite of A's preceeding move, the move is rejected.


I think this would follow the rules exactly.


I think it's correct. It's important to do this for both players on all moves. E.g. a move by player A can both end the mark for player B (if A's move doesn't evade B's lst move) but also start the mark for player A (if A's move threatens one of B's pieces).

Vincent de Boer mentions one other subtlety: how do you compare previous positions? He writes (p113-114) that it is important to give each piece a unique identifier (e.g. their starting setup square). Otherwise positions where multiple pieces of the same rank get swapped during a chase would be judged identical by the server (and that would leak information to the players if not all pieces were known). I assume you already have this?


Quote:

The following would not be forbidden by this algorithm and IMHO also not by the rules:

I move my marshall so that it threats your general. You move your marshal so that it threats my general. I move my general away. You move your general away. And i start again by threatening your general. This would be allowed endlessly because none of us is continuously chasing. We always interrupt that because we both threat and both evade. Correct?

Best regards,
Thorsten


Correct, that exact same situation is discussed here as Example 2: https://stradev.netcorner.org/regeln/remise/

Kind regards,
Rein
Back to top
View user's profile Send private message
spion
Gravon Administrator


Joined: 27 Feb 2002
Posts: 750
Location: Koblenz

PostPosted: 15.01.2021 08:05    Post subject: Reply with quote

Hi,

today i did an update of Gravon's server software. The two squares rule now works correctly and in the original example, a move of the scout from A7 to A5 now is allowed.

Best regards,
Thorsten
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Display posts from previous:   
Post new topic   Reply to topic    Gravon - Das Spielerparadies Forum Index -> General All times are GMT + 1 Hour
Goto page 1, 2  Next
Page 1 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2002 phpBB Group