PDA

View Full Version : Swiss Perfect and pairing errors



Kevin Bonham
15-06-2004, 12:32 AM
from Commonman thread, re NSW Open:


The VIC OPEN had 98 players and started on time for every round. The first round pairings were on the notice-board 45 minutes before the first round scheduled time. As far as I am aware there is no fiddling with the SP pairings. Can some-one expand on what are some inequitable pairings. And if they cause a delay to the schedule are they worth the fiddle?

starter

An example of SP failings:

In the Tas Open SP spat out the following draw for the bottom three boards i round 6:

A Shaw 0 - Hooper 2
C Shaw 1 - Martin 1.5
E Frame 1 - Berry 1

Hooper and Martin could not be paired because there was no legal pairing of the remaining players. C Shaw had not played E Frame and A Shaw had not played Berry but C Shaw and E Frame had both just had two blacks in a row. So far so good.

However Hooper had not played C Shaw and Martin had not played A Shaw. Clearly this pairing:

C Shaw 1 - Hooper 2
A Shaw 0 - Martin 1.5
E Frame 1 - Berry 1

is more equitable than the one SP gives.

I overrode it manually (which in this case only took a few minutes including working out all the above.)

ursogr8
15-06-2004, 08:10 AM
from Commonman thread, re NSW Open:



An example of SP failings:

In the Tas Open SP spat out the following draw for the bottom three boards i round 6:

A Shaw 0 - Hooper 2
C Shaw 1 - Martin 1.5
E Frame 1 - Berry 1

Hooper and Martin could not be paired because there was no legal pairing of the remaining players. C Shaw had not played E Frame and A Shaw had not played Berry but C Shaw and E Frame had both just had two blacks in a row. So far so good.

However Hooper had not played C Shaw and Martin had not played A Shaw. Clearly this pairing:

C Shaw 1 - Hooper 2
A Shaw 0 - Martin 1.5
E Frame 1 - Berry 1

is more equitable than the one SP gives.

I overrode it manually (which in this case only took a few minutes including working out all the above.)

Thanks KB.

But I really had in mind a big field, not 6 players.
Is there an example of where fiddling is required if the field size is 50+.? Does SP have weaknesses for large fields?
Anyone?

starter

Garvinator
15-06-2004, 08:52 AM
Thanks KB.

But I really had in mind a big field, not 6 players.
Is there an example of where fiddling is required if the field size is 50+.? Does SP have weaknesses for large fields?
Anyone?

starter
the situation regarding nsw open round 7 involved at least 10 or more players as charles was working out alexmdc pairing and he was on board eight (i think). The only board that was right for the top eight was board one with Ian Rogers.

ursogr8
16-06-2004, 01:50 PM
the situation regarding nsw open round 7 involved at least 10 or more players as charles was working out alexmdc pairing and he was on board eight (i think). The only board that was right for the top eight was board one with Ian Rogers.

g'ar'g'ar

Thanks for your post even though it does contain alarming news for those of us who just press p for pair then p for print, and let SP take over the pairings.
Do you have any more detail on the incorrectness?

starter

ps Differential is now under 200.

Garvinator
16-06-2004, 03:13 PM
g'ar'g'ar

Thanks for your post even though it does contain alarming news for those of us who just press p for pair then p for print, and let SP take over the pairings.
Do you have any more detail on the incorrectness?


nope but i do intend to send charles an email as i have the sp files now.


ps Differential is now under 200.

shouldnt you be posting post differentials with arosar as he is the next one on your list, not me anymore :p

ursogr8
16-06-2004, 04:05 PM
nope but i do intend to send charles an email as i have the sp files now.



shouldnt you be posting post differentials with arosar as he is the next one on your list, not me anymore :p

I have an undertanding with Amiel that he is going to act as a pacemaker in the pursuit, and then I shoot passed both at the psychological moment. :uhoh:

Kevin Bonham
16-06-2004, 04:33 PM
the situation regarding nsw open round 7 involved at least 10 or more players as charles was working out alexmdc pairing and he was on board eight (i think). The only board that was right for the top eight was board one with Ian Rogers.

It sounds like that situation could have been influenced by modifying the draw to acheive FIDE rating oriented pairing objectives (although I am not certain from what has been said). Although some find this scandalous it is actually not uncommon practice, and is OK provided the distortions are not too great (eg you would not want to float players, or pair the top two in a very large scoregroup.) If there was shuffling for FIDE objectives this is not the fault of Swiss Perfect, but it's not clear from what has been said.

The kinds of SP errors mentioned above are most likely in small fields where forming pairings within score groups can be difficult, leading to batched score groups under the interpretations used in writing the program. When a batched score group includes three or more score groups with none of those score groups making up half or more of the batched score group, silly pairings like apparent needless double-downfloats often arise. I override at least one pairing in the last round of events with 25-30 players and 6-7 rounds about 60% of the time so I'd expect that these problems would sometimes happen with even 50-100 players - but less often.

Garvinator
16-06-2004, 04:38 PM
kevin, as i said i dont know the details as to what charles was doing, i was mainly was trying to let him have some peace while doing the pairings. i do clearly remember him groaning straight away when he noticed the round seven pairings then he started to try and re pair them. I asked him if he had just gotten another example of swiss imperfect at it again and he said yes.

Kevin Bonham
16-06-2004, 05:26 PM
OK, it sounds from that like it was an SP problem and not a FIDE fiddle thing. I have had them at the top end of the field too, but more commonly at the bottom end. There was a rather large dispute in the 2001 June Weekender here after it double-upfloated someone onto top board in the last round and some idiot (me) didn't check it properly and mistakenly thought it was inevitable. We now have a policy that no SP generated draw is official until 10 mins before the start of the next round, even if it has been displayed.