Is the Swiss Pairing System Fair?

Discuss anything you like about chess related matters in this forum.
Nick Pert
Posts: 16
Joined: Wed Jan 26, 2011 6:26 pm

Is the Swiss Pairing System Fair?

Post by Nick Pert » Sat Oct 11, 2014 9:13 am

Over the last few tournaments that I have played I have started to seriously consider whether the Swiss pairing system is actually fair. At the top end of the tournament with a few major contenders it seems to me that the higher the seed you are, the harder the pairings you get on average.

During the early rounds seed 1 gets slightly harder opponents than 2, who in turn gets harder opponents than 3 etc. If you compare seed 1 with seed 3 for example (as they start with the same colour) 1 almost always has harder opponents than 3. The only possible exception would seem to arise if 3 plays 2 before 1 does, but this is not guaranteed to happen and 1 and 2 often play anyway.

The problem can be exaggerated if for example there are 6 players on a score group and one player is significantly lower rated than the others. 1 will almost never get the lower rated player whereas 3 has a much bigger chance.

Then when it comes to floats, if someone is on good form, it is often the job of seed 1 to float up and stop that player, whereas the easy downfloat pairing will almost certainly go to a lower rated player on the same score group.

After thinking of my own issues with the system I started recalling many situations where lower seeded players also had problems with the pairing systems. Two players may be competing for junior, rating or girls prizes or simply against a rival. When one player gets much harder pairings than their rival it can seem unfair.

It seems to me that a much fairer system would be one which takes into account the rating of players that you have already played. So if you have played a high rating average you should be somehow due an easier pairing than a rival who has played a low rating average. I'm not sure exactly how all the details would work but I'm sure someone much cleverer than me can fill in the gaps.

Do we just accept Swiss pairings because we always have? Everyone has sophisticated computer pairing programs these days so maybe it is time to make use of these in providing us with a fairer pairing system!

I’m sure some people disagree with me but thanks for taking the time to read this post anyway!

User avatar
IM Jack Rudd
Posts: 4818
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: Is the Swiss Pairing System Fair?

Post by IM Jack Rudd » Sat Oct 11, 2014 10:00 am

There are Swiss pairing systems that take into account one's field and give the lowest-rated opponents to the players who've faced the highest fields - I think Dubov pairings work that way. I've never tried using them in a tournament.

User avatar
Michael Farthing
Posts: 2069
Joined: Fri Apr 04, 2014 1:28 pm
Location: Morecambe, Europe

Re: Is the Swiss Pairing System Fair?

Post by Michael Farthing » Sat Oct 11, 2014 10:08 am

The Swiss system was only ever designed to find a winner: lower placings are not very reliable. That's a strong argument against using a competition to find a secondary champion (such as Women's Champion or Junior Champion). The justification for the winner depends on their being sufficient rounds so that the top contenders will all have faced each other - that effectively means about 10 rounds for a field of 32.

On Nick's second point I seem to recall that before gradings or ratings were as prominent as nowadays, pairings after round 1 did use progressive sum of opponent's scores as a means of evening up the strength of opposition players on the same score had faced. Can anyone give more detail? Why was this sort of objective discarded?

Jonathan Bryant
Posts: 3452
Joined: Sun May 11, 2008 3:54 pm

Re: Is the Swiss Pairing System Fair?

Post by Jonathan Bryant » Sat Oct 11, 2014 10:23 am

Nick Pert wrote:... At the top end of the tournament with a few major contenders it seems to me that the higher the seed you are, the harder the pairings you get on average. ...

Two players may be competing for junior, rating or girls prizes or simply against a rival. When one player gets much harder pairings than their rival it can seem unfair.

(Unfortunately) I rarely get to experience your first problem.

With regard to your second example - yes, for sure swiss pairings are not a good way to allocate prizes for people finishing in the middle of the field. I know some events give grading prizes based on TPR rather than absolute score, although that does add a layer of complexity which is not necessarily a good idea. There’s a value in simplicity, I’d say.

Another example of 'unfair' swiss results ...

It seems to me that often in the British there’s somebody who ends up getting a decent prize - a share of third, say - without playing many (or even any) of the leading contenders. Half a point below them, going home with little or nothing, are higher rated opponents who’ve played most of their games against the toughest opponents.

Not the best system. Perhaps it's the best we can do, though.

Alex Holowczak
Posts: 9085
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire

Re: Is the Swiss Pairing System Fair?

Post by Alex Holowczak » Sat Oct 11, 2014 8:49 pm

Nick Pert wrote:Do we just accept Swiss pairings because we always have? Everyone has sophisticated computer pairing programs these days so maybe it is time to make use of these in providing us with a fairer pairing system!
Far from it. In fact, British arbiters are usually the ones who complain about the FIDE Pairing Rules (i.e. the Dutch system), and instead have created and used their own Pairing Rules, because they believe them to be fairer. I understand they are under constant review. These are in use at the British.

stevencarr

Re: Is the Swiss Pairing System Fair?

Post by stevencarr » Sat Oct 11, 2014 11:57 pm

Is the Swiss system fair?

When was the last Swiss tournament where somebody could honestly say it was not fair for the eventual winner to have actually won that competition?

This is not a rhetorical question. Does anybody know of such instances?

Nick Pert
Posts: 16
Joined: Wed Jan 26, 2011 6:26 pm

Re: Is the Swiss Pairing System Fair?

Post by Nick Pert » Sun Oct 12, 2014 10:05 pm

I'm not trying to say that anyone didn't deserve to win a tournament. You can only play what is in front of you. There are however numerous examples where 2 players get the same score but one has played significantly higher rated players than another. Players can also get higher scores than other players despite much lower performances. All I'm trying to say is that if a player has had a tough run of pairings it seems right that they should get some kind of credit for that in future rounds.

Paul Bielby
Posts: 154
Joined: Tue Dec 22, 2009 4:14 pm
Location: South Shields

Re: Is the Swiss Pairing System Fair?

Post by Paul Bielby » Sun Oct 12, 2014 10:56 pm

Hi Nick,

This reminds me of an argument I had with Stewart Reuben when I first became an arbiter 30 odd years ago. It all depends on how you define fair.

The present system is obviously weighted in favour of the top rated players as you yourself have pointed out. The number one seed is given the easiest path to the top prize. just as the driver in pole position has the best chance in an F1 race. It's designed to be a self-fulflling prophesy.

My contention in the argunment with Stewart was that if all entrants are to have an equal chance in the draw (my definition of 'fair') then the only fair system was for draw in each round to be entirely random. Of course it then becomes possible for the nos 1 and 2 seeds to have to play each other in R1. But it is fair that this is just as possible as any other pairing. All right, so that's 'fair', but is it acceptable? Or indeed even practical (The draw for each round would need to be performed publicly using some sort of lottery system to be seen to be fair) The present system doea have the virtue that the protocol laid down means that players and arbiters following it will all reach the same draw.

I can remember once, in an earlier Scarborough Congress in the McGregor era, random pairings were used. The top two seeds (was it Mark Hebden and Dave Mooney?) were indeed paired to play in R2. They hated it and I don't blame them- sure it was entirely 'fair' but surely not acceotable.

User avatar
David Shepherd
Posts: 912
Joined: Fri Nov 23, 2007 3:46 pm

Re: Is the Swiss Pairing System Fair?

Post by David Shepherd » Sun Oct 12, 2014 11:05 pm

Paul Bielby wrote:Hi Nick,
The present system is obviously weighted in favour of the top rated players as you yourself have pointed out. The number one seed is given the easiest path to the top prize. just as the driver in pole position has the best chance in an F1 race. It's designed to be a self-fulflling prophesy.
I understood Nick to be arguing the reverse of the above i.e. the top seed often gets the hardest draw out of the top players (in a Swiss using normal pairings with a low number of rounds)

User avatar
IM Jack Rudd
Posts: 4818
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: Is the Swiss Pairing System Fair?

Post by IM Jack Rudd » Sun Oct 12, 2014 11:09 pm

Here is a good example of the kind of the oddities the Swiss system can throw up. Spot the runner-up who probably didn't deserve to finish that high. :D

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: Is the Swiss Pairing System Fair?

Post by Roger de Coverly » Sun Oct 12, 2014 11:19 pm

David Shepherd wrote: I understood Nick to be arguing the reverse of the above i.e. the top seed often gets the hardest draw out of the top players (in a Swiss using normal pairings with a low number of rounds)
It may not always be the number one seed, but the tournament leader can also get demonstrably stronger opposition than the second place. It's usually only apparent in long tournaments such as the British Championships.

User avatar
Rob Thompson
Posts: 757
Joined: Wed Jul 15, 2009 12:03 pm
Location: Behind you

Re: Is the Swiss Pairing System Fair?

Post by Rob Thompson » Sun Oct 12, 2014 11:20 pm

I guess the top seed has the "advantage" of not having to play the top seed.
True glory lies in doing what deserves to be written; in writing what deserves to be read.

User avatar
Michael Farthing
Posts: 2069
Joined: Fri Apr 04, 2014 1:28 pm
Location: Morecambe, Europe

Re: Is the Swiss Pairing System Fair?

Post by Michael Farthing » Mon Oct 13, 2014 9:05 am

On the contrary, Rob, were he allowed to do so he would be guaranteed a full point.

Michael Flatt
Posts: 1235
Joined: Tue Jul 02, 2013 7:36 am
Location: Hertfordshire

Re: Is the Swiss Pairing System Fair?

Post by Michael Flatt » Mon Oct 13, 2014 9:16 am

The Swiss Pairing System is designed to identify a winner or small number of joint winners from a large field of players. For it to achieve its aims all players are ranked (seeded) according to playing strength so that the top seeds do not meet until the later rounds when the top score group is small in number. The major benefit is that all players play in all rounds unlike a knock-out when the field is halved in each round. In a knock-out all games must be decisive. In a Swiss Tournament it is still possible to win the tournament even if some games are lost or drawn.

In an all-play-all tournament the field has be restricted due to the practical constraint on the number of rounds that can be played.

Perhaps, a hybrid tournament might be considered fairer where a Swiss tournament is run to identify a small field of qualifiers who could then play in an all-play-all.

The alternative is to play in a stronger tournament where you are not the top seed!

Ian Thompson
Posts: 3551
Joined: Wed Jul 02, 2008 4:31 pm
Location: Awbridge, Hampshire

Re: Is the Swiss Pairing System Fair?

Post by Ian Thompson » Mon Oct 13, 2014 6:37 pm

Michael Flatt wrote:The alternative is to play in a stronger tournament where you are not the top seed!
Not that easy to do though when you're ranked 7th in the country. :)