Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- From: jazzerciser@xxxxxxxxxxx (-)
- Date: Tue, 06 Sep 2005 08:13:02 GMT
> jazzerciser@xxxxxxxxxxx (-) wrote:
>> Europeans can't seem to operate a tournament properly.
Robert Jasiek <jasiek@xxxxxxxx> wrote:
> Do you draw general conclusions from the one particular tournament
> TT Brno?
I believe that I am adding fuel to the fires of the European Open 2005.
>> Note how Mr. Cornel Burzo was paired against a 3d in Round #3
>> after losing one game !
> He, #3, was paired against player #2, a 7d, in round 3 and against
> player #16, a 3d, in round 4.
Yes, I meant to say that Cornel was paired against a 3d in Round #4.
> After round 3, the following players had these MMS:
>
> final MMS in top
> place round 3 group
>
> #1 26 x
> #2 26 x
> #3 25 x
> #4 25 x
> #5 25 x
> #6 25 x
> #7 25 x
> #8 25 x
> #9 24 x
> #10 25 x
> #11 24 x
> #12 24 x
> #13 24 x
> #14 24 x
> #15 24 x
> #16 24
> #17 23 x
> ... <24
>
> For round 4, this forces the game #1-#2. The even number of players
> #3, #4, #5, #6, #7. #8 are available for pairing next, unless all
> possible pairings would repair an earlier game of the tournament.
> Since the pairing #3-#4, #5-#6, #7-#8 is one such a possibility, all
> these MMS25 players have to be paired against each other in round 4.
Well, the MMS25 players -never- encountered #3-#4, #5-#6, #7-#8.
Doesn't anybody in Europe find that rather odd? Did nobody examine
the pairings, during the tournament, or comment after the tournament?
Why was Christian Pop. given an easy game in Round #2 ?
> That #3-#16 was a game in round 4 does not conform to what one expects
> from a MacMahon tournament, although "MacMahon" has never been
> defined officially so far. Thank you for pointing out this!
One may infer that 2^4 = 16 generated a breakpoint down to #16
(though prior to the assignment of final standings). This was reaching
too far, as the pairing algorithm allowed a 3 rank difference too early.
A proper pairing algorithm would have rejected Round #4 immediately.
>> When Mr. Cho. Seok Bin lost a game he
>> was paired against a 6d in Round #5.
> ... In round 5, Cho Seok-bin, #2, might have been paired against one
> of #3, #4, #5, #6, if possible. That he was paired against #4 was ok.
Yes, I am observing the disparity. After losing one game, Mr. Cho.
Seok Bin was paired against a player with one rank difference, unlike
Mr. Burzo who was paired against a player -THREE- ranks apart !!!
> (There is no specification for EGF tournaments of how a pairing must
> be done if MMS are equal and several opponents are thus available.)
Examination of the entire pairing space can arrive at most likely
course toward maximum likelihood. This approach was applied for
the first time in the pairing algorithms for the U.S. Open 2005.
>> When Radel Nechanicky
>> lost a game he was paired against a 4d.
> He was paired down in round 6.
Mr. Nechanicky was also paired down in Round #4. Mr. Vit Brunner
is -TWO- ranks different in previously declared playing strength. The
match with Mero Csaba never occurred, though could have been made.
>> The problem here concerns
>> heavy reliance (incorrect) on the SOS and SOSOS formulations.
> Why? Do you know which pairing program was used and with which
> parameters? Before one cannot say whether the pairing depended on
> tiebreakers at all.
SOS and SOSOS are not proper methods for tiebreakers. The
criteria for good tiebreaker is this: interim pairing methods must not
be any different from end-round tournament tiebreakers. A tournament
is simply an aborted ratings procedure. You will find a consistent
application by using SODOS, SODOSODOS, SODOSODOSODOS ...
>> The column titled "MMS" is not actually MacMahon Score, but
>> SOS.
> Now you are joking.
Am I ? You said "MacMahon" has never been officially defined.
Here MMS seems to substitute for an internal ranking method. Who
can say how the MMS column was formulated without an explanation?
>> When enough players complain then tournament organizers
>> will need to justify what they are doing upon mathematical grounds,
> Rather the EGF Rules Commission would disallow pairing programs
> that produce non-MacMahon pairings.
This Toyota-PandaNet Tour Brno 2005 used non-MacMahon pairings.
>> Players are not expected to know mathematics, however the
>> tournament organizers who operate pairing programs should be
>> accountable to the mathematical arguments and proofs.
> As far as it is clear what MacMahon as a system is supposed to do.
There is no conflict between maximum likelihood and MacMahon.
>> Robert Jasiek is perhaps responsible for the European travesties
>> today,
>
> No, but because I am a member of the EGF Rules Commission.
Does the EGF Rules Commission sanction the results of this
Toyota-PandaNet Tour Brno 2005 ?
>> because Mr. Jasiek has incorrectly argued for use of SOS rather
>> than counting up the defeat chains in SODOSODOS and so forth
>> (getting longer for each successive round).
> I have not argued such for producing pairings yet. I have argued
> about the final results and favoured MMS only without tiebreakers
> for the final results.
I believe that your periodic FAQ mentions SOS but does not
examine the utility of SODOSODOS -> SODOSODOSODOS ... etc.
MacMahon Score is discussed only because this is what people
seem capable of understanding. Ultimately, maximum likelihood
methods will be among the solutions that resolve these problems.
- regards
- jb
----------------------------------------------------------------------
(Jewish) Judge all washed up
http://www.nationalvanguard.org/story.php?id=5995
----------------------------------------------------------------------
.
- Follow-Ups:
- Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- From: Robert Jasiek
- Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- References:
- Prev by Date: Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- Next by Date: Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- Previous by thread: Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- Next by thread: Re: Traviesties in the methods Europeans Use for Pairing Algorithms
- Index(es):
Relevant Pages
|
