PDA

View Full Version : Protos



Brian_Jones
26-10-2005, 04:03 PM
Why does Australia not use the FIDE approved German pairing program PROTOS? It is free and produces better parings than the Australian progam Swiss Perfect, which is no longer being maintained. It is not too late to adopt PROTOS and interface with ACF rating system.

Bill Gletsos
26-10-2005, 04:06 PM
Why does Australia not use the FIDE approved German pairing program PROTOS? It is free and produces better parings than the Australian progam Swiss Perfect, which is no longer being maintained. It is not too late to adopt PROTOS and interface with ACF rating system.Last I looked PROTOS was a DOS based program and was not very user friendly.

Ian Rout
26-10-2005, 10:08 PM
Last I looked PROTOS was a DOS based program and was not very user friendly.
I downloaded it to have a look about a month ago, noting that it was proclaimed as a new version, but didn't get very far through the user-hostile interface or the manual so I can't comment on how well it does the pairings. I'll have to take Brian's word on that.

In its current form I would recommend it to lovers of late-1970s nostalgia; perhaps it could be relaunched as Polyester.

themovingman
27-10-2005, 01:05 AM
now I haven't run it

I may have even downloaded the wrong one

the above given : talk about SwissP not being maintained - the most recent date of this package is 2001 [not too bad of course and it may not need to be maintained but if we're gonna talk about maintained !]

And Yes is a DOS based thing [again don't have to be a black mark against anything], perhaps once you use it several times things go really smoothly, what are the shape of the reports - for rating; putting up a web/email report.

If anyone has experience of these, I will have a bash on an already existing tourament but a) my deadlines always slip and b) someone may know and type the answer within minutes !

Denis_Jessop
27-10-2005, 12:58 PM
I downloaded it to have a look about a month ago, noting that it was proclaimed as a new version, but didn't get very far through the user-hostile interface or the manual so I can't comment on how well it does the pairings. I'll have to take Brian's word on that.

In its current form I would recommend it to lovers of late-1970s nostalgia; perhaps it could be relaunched as Polyester.

So did I, Ian, and with almost the same reaction though I didn't even get as far as you did.

As for SP not being maintained (another point) I assume League Watch will be up-to-date though I haven't tried it and the ACF has no deal regarding it so far. As the Swiss Rules haven't changed since the last up-date it's only the alleged bugs that may need fixing though I understand that Robert Rosycki doesn't concede that there are any. Has anyone compared LW's pairings with SP's?

DJ

Bill Gletsos
27-10-2005, 01:31 PM
So did I, Ian, and with almost the same reaction though I didn't even get as far as you did.

As for SP not being maintained (another point) I assume League Watch will be up-to-date though I haven't tried it and the ACF has no deal regarding it so far. As the Swiss Rules haven't changed since the last up-date it's only the alleged bugs that may need fixing though I understand that Robert Rosycki doesn't concede that there are any. Has anyone compared LW's pairings with SP's?Last time I looked at it (last couple of months), Leaguewatch had a number of issues. e.g. I am yet to find a way to show the players id number in any view/list. You cannot import players from .trn files.

pax
27-10-2005, 05:57 PM
It might be more accurate than SP, but you'd have to be out of your mind to want to use it -it's quite diabolical.

erral
08-12-2005, 05:02 AM
Has anyone compared LW's pairings with SP's?

DJ

I haven't tested it, but a friend told me that LW uses the same pairing procedure as SP.

The best Swiss Pairing programs for me are Swiss Manager and Swiss Master (Petunia).

Regards,

Mikel Larreategi
International Arbiter

Garvinator
05-01-2006, 09:19 AM
Gary Bekker and I set up protos for the australian championship yesterday to see what pairings etc it would give as opposed to sp and I cant recommend protos to be used for any type of large tournament.

While the pairings seemed to be ok at a quick glance, it is very unweildy and way too difficult to use when time pressured like arbiters generally are in large weekenders.

This is one of the big advantages of sp, it is relatively easy to use and is easily viewable. The small errors it makes can be easily rectified.

jase
05-01-2006, 05:24 PM
Why does Australia not use the FIDE approved German pairing program PROTOS? It is free and produces better parings than the Australian progam Swiss Perfect which is no longer being maintained.

On what basis do you make this claim?
My impressions from these Australian Championships is that the PROTOS pairings are unsound.

jenni
05-01-2006, 05:43 PM
I have been told that there is a pairing program (and I think it might be swiss manager although I am not sure) that produces correct pairings. Cathy Rogers said it was used in Malaysia and Manuel Weeks says there is a correct one used overseas.

It seems insane to me that we use a pairing program in Australia that requires DOP's to constantly manually intervene. As the draws get more complicated it takes longer and longer for the draws to be produced - e.g the girls under 18 draw took 1/2 hour to produce today, because changes and checking was required. Surely it is better to have software that can be trusted?

I'll have to have a look on the net to see what the program is.

shaun
05-01-2006, 05:59 PM
I have been told that there is a pairing program (and I think it might be swiss manager although I am not sure) that produces correct pairings. Cathy Rogers said it was used in Malaysia and Manuel Weeks says there is a correct one used overseas.

It seems insane to me that we use a pairing program in Australia that requires DOP's to constantly manually intervene. As the draws get more complicated it takes longer and longer for the draws to be produced - e.g the girls under 18 draw took 1/2 hour to produce today, because changes and checking was required. Surely it is better to have software that can be trusted?

I'll have to have a look on the net to see what the program is.

I'm pretty sure that it is Petunia (Swiss Master) that Cathy was talking about. I have run the demo version and find it harder to use than Swiss Perfect.
As for the topic of "correctness", this is (as you know) a constant source of disagreement amongst arbiters in Australia. My position is that Swiss Perfect does the pairings according to the FIDE Swiss Pairing Rules, and if you have problems with arbiters manually intervening with the draw you need to ask them on what grounds they are doing so. (Make sure you get them to show you the EXACT rule that Swiss Perfect has broken).

Bill Gletsos
05-01-2006, 06:00 PM
On what basis do you make this claim?
My impressions from these Australian Championships is that the PROTOS pairings are unsound.It is my understanding that all FIDE approved swiss pairing programs have to correctly pair a set of tournaments created by the FIDE Swiss Pairing Commission to gain approval. As such one would assume that the FIDE approved PROTOS does the pairings in the same manner as the other FIDE approved programs SWISS Master 5.0 from The Netherlands and SWISS Manager from Austria. Of course it could always be possible that the programs only get the same results for the test set of tournaments.

BTW I assume you are aware that PROTOS is written by the Chairman of the FIDE Swiss Pairing Commission Christian Krause. You would think he would know how to apply the FIDE swiss pairing rules, however maybe he doesnt.

jenni
06-01-2006, 09:33 AM
I'm pretty sure that it is Petunia (Swiss Master) that Cathy was talking about. I have run the demo version and find it harder to use than Swiss Perfect.
As for the topic of "correctness", this is (as you know) a constant source of disagreement amongst arbiters in Australia. My position is that Swiss Perfect does the pairings according to the FIDE Swiss Pairing Rules, and if you have problems with arbiters manually intervening with the draw you need to ask them on what grounds they are doing so. (Make sure you get them to show you the EXACT rule that Swiss Perfect has broken).

OK I have had a look on the net and from what Manuel told me about where to find it and what Cathy told me last night, yes I think they are talking about Swiss Master.

I think the problem is that for a very high proportion of "top" tournaments now manual changes are happening. I know Charles Z does it in every tournament he Dops (and that is a lot). You don't even know it is happening, so really there is not the opportunity to challenge them. and I wouldn't anyway, although having gone to an arbiters workshop and bought the book, I am busy learning all the rules. It seems to me that draws can become more of an art than a science.

As long as arbiters are unhappy with the software and insisting on changing the draw then surely we need to start using software that they are happy with. One of the rounds of the championships was a debacle because the incorrect pairings were put up. That seems to have been because the altered pairings somehow got lost.

If we used a program that everyone was happy with, then maybe we could still use swiss perfect for clubs and schools, but make the other one the tournament standard? And then the ACF could have a rule that no manual intervention was allowed. Seems to me it would make everyone's life a lot easier?

Of course I have no idea of the ramifications as far as processing ratings goes - maybe Bill could tell us?

shaun
06-01-2006, 10:18 AM
I think the problem is that for a very high proportion of "top" tournaments now manual changes are happening. I know Charles Z does it in every tournament he Dops (and that is a lot). You don't even know it is happening, so really there is not the opportunity to challenge them. and I wouldn't anyway, although having gone to an arbiters workshop and bought the book, I am busy learning all the rules. It seems to me that draws can become more of an art than a science.


Arbiters should do two things concerning swiss pairings. (A) Post in the tournament hall (or at least make available) the set of pairing rules they are using. (B) Be able to explain how the draw was done, with specific reference to the posted rules, even if the pairings were done by a cimputer program.
Any player is entitled to ask for an explanation about how the pairings were arrivied at (as long as they ask nicely).



As long as arbiters are unhappy with the software and insisting on changing the draw then surely we need to start using software that they are happy with. One of the rounds of the championships was a debacle because the incorrect pairings were put up. That seems to have been because the altered pairings somehow got lost.

If we used a program that everyone was happy with, then maybe we could still use swiss perfect for clubs and schools, but make the other one the tournament standard? And then the ACF could have a rule that no manual intervention was allowed. Seems to me it would make everyone's life a lot easier?

I for one am happy with using Swiss Perfect in the tournaments I direct, simply because I believe that Swiss Perfect follows the FIDE Pairing Rules, as they are written. And I would be happy to see the ACF make a rule that no manual intervention was allowed, except in the case where it could be CLEARLY demonstrated that a SPECIFIC Fide rule was broken.

jenni
06-01-2006, 10:35 AM
I for one am happy with using Swiss Perfect in the tournaments I direct, simply because I believe that Swiss Perfect follows the FIDE Pairing Rules, as they are written. And I would be happy to see the ACF make a rule that no manual intervention was allowed, except in the case where it could be CLEARLY demonstrated that a SPECIFIC Fide rule was broken.
Maybe DOPs should have to write a report to the ACF anytime they change a pairing, with full justification.

Most people hate paperwork and it would make them think twice about it....

Oepty
06-01-2006, 10:51 AM
Maybe DOPs should have to write a report to the ACF anytime they change a pairing, with full justification.

Most people hate paperwork and it would make them think twice about it....

Jenni, that would be very good, except what one arbiter thinks is justification another arbiter thinks is a load of nonsense. How then are you going to judge the justification? Who is going to judge the justification? If the ACF accepted a justification as being correct they then are effectively saying they believe one interpretation of the rules is correct.

Scott

Bill Gletsos
06-01-2006, 11:26 AM
Protos, Swiss Master (previously Petunia) and Swiss Manager are all FIDE Swiss Pairing Committee approved. There should be virtually no reason for an arbiter to manual change the pairings generated by these programs. Certainly if one was to do so I would think they should then justify it by refering it to the FIDE Swiss Pairing Committee.

Oepty
06-01-2006, 12:25 PM
Protos, Swiss Master (previously Petunia) and Swiss Manager are all FIDE Swiss Pairing Committee approved. There should be virtually no reason for an arbiter to manual change the pairings generated by these programs. Certainly if one was to do so I would think they should then justify it by refering it to the FIDE Swiss Pairing Committee.

I think that seeing there is a difference of opinion within Australian IA's then some approach should be made to FIDE Swiss Pairing Committee to ask them to made a public clarification of what is the correct pairing in the situations. Scott

Garvinator
06-01-2006, 12:49 PM
I think that seeing there is a difference of opinion within Australian IA's then some approach should be made to FIDE Swiss Pairing Committee to ask them to made a public clarification of what is the correct pairing in the situations. Scott
now this would be good. Even better would be an update to the dutch pairing rules to make them clearer as to whether top v bottom or colour is more important. Also sometimes whether floats or colour is more important can be an issue and doesnt seem clear from reading the pairing rules.

Garvinator
06-01-2006, 12:51 PM
I think that seeing there is a difference of opinion within Australian IA's then some approach should be made to FIDE Swiss Pairing Committee to ask them to made a public clarification of what is the correct pairing in the situations. Scott
Part of the issue is when the arbiters were first learning to pair players and what rules they were learning under. Some learnt under the old lim pairing rules, therefore have carried forward some of the pairing rules from that time to now, as parts of the old rules are still incorporated into the dutch pairing rules.

pax
06-01-2006, 02:15 PM
now this would be good. Even better would be an update to the dutch pairing rules to make them clearer as to whether top v bottom or colour is more important. Also sometimes whether floats or colour is more important can be an issue and doesnt seem clear from reading the pairing rules.

Hmm, after reading the FIDE handbook on this, and the old top/bottom vs colour discussion (http://www.chesschat.org/showthread.php?t=2214) I wonder whether the wording has in fact changed.

In Shaun's post on that thread, he refers to B1 (absolute criteria) and B2 (relative criteria). Looking on the FIDE web site, it now labels the absolute criteria (not playing the same player twice, and not recieving the same colour three times or more than +/-2) as B1 and B2 respectively, and the relative criteria as B3 through B6.

Now when you read that in conjunction with this:


C.5
Order the players in S1 and S2 according to A2.
C.6
Pair the highest player of S1 against the highest one of S2, the second highest one of S1 against the second highest on e of S2, etc. If now p pairings are obtained in compliance with B1 and B2 the pairing of this score bracket is considered complete.

It seems clear: you pair top vs bottom, and if no absolute preferences are violated the pairing is complete.

The trouble then, is that the pairing procedures never mention the colour preference clause B4!! Why is the rule there if it is never to be used??

pax
06-01-2006, 02:18 PM
What FIDE really ought to do is release the source code for PROTOS, or at least a proper algorithmic procedure for pairing. Then we could all have an unambiguous reference for how to correctly perform the pairings. Not only that, but we could turn PROTOS into something that is actually usable!

Bill Gletsos
06-01-2006, 03:36 PM
Part of the issue is when the arbiters were first learning to pair players and what rules they were learning under. Some learnt under the old lim pairing rules, therefore have carried forward some of the pairing rules from that time to now, as parts of the old rules are still incorporated into the dutch pairing rules.As I understand it the Lim rules still exist and are entirely seperate from the Dutch rules.

Check http://www.fide.com/official/handbook.asp?level=C04

04.1 is the Dutch rules and 04.2 is the Lim rules.

As Geurt noted in one of his notebook's

In the FIDE Handbook you can find the systems and the programs,
which are approved by FIDE:
There are four systems approved by FIDE:
1. Lim System: this is a Swiss system based on ratings. There is a
computer program developed by GMB.
2. Dutch System: this system is also based on ratings. There are several
computer programs to make pairings with this system: Petunia, Swiss
Chess, SVBOSS
3. Dubov system: this system is also based on ratings, but uses the TPR’s
achieved after each round. As far as I know there is no computer
program for the Dubov system.
4. Burstein system: this system is based on the Buchholz scores of the
players after each round. I do not know a computer program that is
able to make pairings according to this system.

The systems are described in Chapter C of the FIDE Handbook, except the
Burstein system. An explanation of the Burstein system is given in Chapter D
(see Olympiad Pairings Rules).

Bill Gletsos
06-01-2006, 03:55 PM
Stewart Reubens notes on page 34 of the Chess Organiser's Handbook Second Edition that "I have never heard comments about the Petunia Program for the Dutch System giving the wrong pairings".

An intersting point he makes on the same page is "They are all based on the premise that throughout the tournament all players on all score groups are paired top half v bpottom half in rating order after taking into account colour equalisation and then colour alternation." (the bolding of after is mine)

Petunia/Swiss Master is also the program the Geurt swears by.

It would be interesting to compare the draw done for the Doeberl by SP and by Swiss Master 5.1.

N.B. On the FIDE web site under the section on Swiss rules, Regualtions for swiss tournaments, Miscellaneous it notes "The FIDE Swiss Rules pair the players in an objective and impartial way, and different arbiters following the pairing rules should arrive at identical pairings. "

Bill Gletsos
06-01-2006, 04:03 PM
Hmm, after reading the FIDE handbook on this, and the old top/bottom vs colour discussion (http://www.chesschat.org/showthread.php?t=2214) I wonder whether the wording has in fact changed.

In Shaun's post on that thread, he refers to B1 (absolute criteria) and B2 (relative criteria). Looking on the FIDE web site, it now labels the absolute criteria (not playing the same player twice, and not recieving the same colour three times or more than +/-2) as B1 and B2 respectively, and the relative criteria as B3 through B6.

Now when you read that in conjunction with this:


C.5
Order the players in S1 and S2 according to A2.
C.6
Pair the highest player of S1 against the highest one of S2, the second highest one of S1 against the second highest on e of S2, etc. If now p pairings are obtained in compliance with B1 and B2 the pairing of this score bracket is considered complete.
It seems clear: you pair top vs bottom, and if no absolute preferences are violated the pairing is complete.

The trouble then, is that the pairing procedures never mention the colour preference clause B4!! Why is the rule there if it is never to be used??Although C6 refers to B1 and B2 it really should be referring to B1 to B6. In fact Stewart Reuben tries to rectify this in the section of his book The Chess Organisers Handbook Second edition on the Dutch rules by renumbering B1 & B2 as just B1 with sections (a)-(e) and renumbering B3 to B6 as B2 with sections (a)-(d).

I note however that in his latest addition he has gone back to simply duplicvating exactly what is on the FIDE website.

Garvinator
06-01-2006, 06:43 PM
Hmm, after reading the FIDE handbook on this, and the old top/bottom vs colour discussion (http://www.chesschat.org/showthread.php?t=2214) I wonder whether the wording has in fact changed.

In Shaun's post on that thread, he refers to B1 (absolute criteria) and B2 (relative criteria). Looking on the FIDE web site, it now labels the absolute criteria (not playing the same player twice, and not recieving the same colour three times or more than +/-2) as B1 and B2 respectively, and the relative criteria as B3 through B6.

Now when you read that in conjunction with this:


It seems clear: you pair top vs bottom, and if no absolute preferences are violated the pairing is complete.

The trouble then, is that the pairing procedures never mention the colour preference clause B4!! Why is the rule there if it is never to be used??

This was discussed at length by the four ia over a couple of rounds as to which takes precedence and how to apply it. It was decided by vote that as B4 is mentioned first ie before section C, that it takes precedence.

Thinking further and having done absolutely no testing (so it is just thoughts), I think players in a score group are divided into S1 and S2. Then the top ranked player in S1 is paired with someone from S2 who has the best colour preference and meets the other criteria ie B1 and B2. If all players from S2 want the same preference as player 1, then you look at the players in S1, starting with the lowest ranked player in S1.

If no player can be found that fits with player 1's colour preference, then you pair him with the top player in S2 (if possible) and allocate colours, which will mean that one player gets two of the same colour in a row.

I will have a chat to Jason and Gary tomorrow and see what they think.

shaun
06-01-2006, 06:58 PM
Thinking further and having done absolutely no testing (so it is just thoughts), I think players in a score group are divided into S1 and S2. Then the top ranked player in S1 is paired with someone from S2 who has the best colour preference and meets the other criteria ie B1 and B2. If all players from S2 want the same preference as player 1, then you look at the players in S1, starting with the lowest ranked player in S1.

If no player can be found that fits with player 1's colour preference, then you pair him with the top player in S2 (if possible) and allocate colours, which will mean that one player gets two of the same colour in a row.

I will have a chat to Jason and Gary tomorrow and see what they think.

The following bit is wrong (based on me misreading the above quote)
*****
This is certainly the thinking that a lot of people have concerning the FIDE swiss pairing rules, and one that I think is wrong.
*****

The rest is right, but superfluous
*****
Reading the rules will show that when you reach the situation where no player can be found that fits with player 1's colour preferences, you swap the bottom player in S1 with the top player in S2 and try again. You repeat this process (swapping between s1 and s2 as defines in the rules), until all players in the score group get their colour preferences met. Only if this is impossible do other things happen ....
*****

pax
06-01-2006, 07:43 PM
Thinking further and having done absolutely no testing (so it is just thoughts), I think players in a score group are divided into S1 and S2. Then the top ranked player in S1 is paired with someone from S2 who has the best colour preference and meets the other criteria ie B1 and B2. If all players from S2 want the same preference as player 1, then you look at the players in S1, starting with the lowest ranked player in S1.

If no player can be found that fits with player 1's colour preference, then you pair him with the top player in S2 (if possible) and allocate colours, which will mean that one player gets two of the same colour in a row.




This is certainly the thinking that a lot of people have concerning the FIDE swiss pairing rules, and one that I think is wrong. Reading the rules will show that when you reach the situation where no player can be found that fits with player 1's colour preferences, you swap the bottom player in S1 with the top player in S2 and try again. You repeat this process (swapping between s1 and s2 as defines in the rules), until all players in the score group get their colour preferences met. Only if this is impossible do other things happen ....

Wait a sec. Aren't you effectively saying the same thing here? What am I missing?

jenni
06-01-2006, 07:50 PM
Jenni, that would be very good, except what one arbiter thinks is justification another arbiter thinks is a load of nonsense. How then are you going to judge the justification? Who is going to judge the justification? If the ACF accepted a justification as being correct they then are effectively saying they believe one interpretation of the rules is correct.

Scott

It was a bit of a joke Scott - make it so bureaucratic, no-one would be bothered to do manual changes...

shaun
06-01-2006, 08:04 PM
No. In Garvin's description the players are placed into S1 and S2 at the start of the process, and there are NO exchanges between these groups. In the FIDE rules players can be exchanged between groups during the pairing process, so a player who is the bottom half of a group (eg initially in S2) can be swapped so she is paired as though she is in the top half of the group. This has the effect of evening up the number of colour matches between both groups.
eg S1 consists of 3 players with player 1 needing a W, player 2 W, player 3 B, and S2 consists of 3 players, with player 4 needing B, 5 B and 6 W. If player 1 has played players 4 and 5 then under the top v bottom paradigm player 1 would play player 6 and player 6 would just get another black. The other pairings would be 2 v 4 and 5 v 3 with 5 getting a second white.
Under the FIDE system players 4 and 3 would be exchanged and S1 would now contain players 1,2,4 and S2 would contain 3,5,6 resulting in 1V3 2V5 and 6V4

jenni
06-01-2006, 08:05 PM
Stewart Reubens notes on page 34 of the Chess Organiser's Handbook Second Edition that "I have never heard comments about the Petunia Program for the Dutch System giving the wrong pairings".

An intersting point he makes on the same page is "They are all based on the premise that throughout the tournament all players on all score groups are paired top half v bpottom half in rating order after taking into account colour equalisation and then colour alternation." (the bolding of after is mine)

Petunia/Swiss Master is also the program the Geurt swears by.

It would be interesting to compare the draw done for the Doeberl by SP and by Swiss Master 5.1.

This seems a sensible idea - it seems crazy not to at least try and see if software is available which makes manual pairings unnecessary.


N.B. On the FIDE web site under the section on Swiss rules, Regualtions for swiss tournaments, Miscellaneous it notes "The FIDE Swiss Rules pair the players in an objective and impartial way, and "

hmm - as I said I am getting the impression this is more of an art than a science.

pax
06-01-2006, 08:11 PM
No. In Garvin's description the players are placed into S1 and S2 at the start of the process, and there are NO exchanges between these groups.

But he said if you can't find a partner for player one in S2, then you look in S1 starting from the bottom - effectively swapping that player into S2. Obviously the exchanges described in section D are rather more precise, but still it is the same idea.

shaun
06-01-2006, 08:16 PM
I stand corrected. I missed this bit
If all players from S2 want the same preference as player 1, then you look at the players in S1, starting with the lowest ranked player in S1.