Random Pairings Make Chess Players Go Crazy

At the Chess.com Isle of Man International, which is taking place right now, the entire chess world’s brains exploded when the first round of the Open used a random pairing system, and Caruana and Kramnik got paired.

exploding_head_3

All of this shows me the lack of imagination in the chess world.

How many open chess tournaments do we have each year? Hundreds, maybe thousands even? How many of them use the exact same pairing system? It’s pretty close to all of them.

So we have every single tournament in the world, doing the exact same thing every time. I’m going to admit that the Swiss System is a great system, and works perfectly fine as the standard pairing system for an open event. However the problem with the chess community is:

Almost any change causes everyone to freak out

Is a random pairing system unfair? Absolutely not. The results of the random draw for one particular event may be unfair, but over the long run it will turn out to be pretty fair. Everyone worries so much about Carlsen, Kramnik and how important their pairings are, but there are hundreds of players in the tournament and they matter too.

If there are 100 players, why should player 51 automatically have to play the top seeded player when nearly half the field is lower rated than her?

Why should someone ranked 70th have to play a tougher opponent than someone ranked 100th? Is that objectively fair?

This is not a closed event, and I think it’s reasonable for every player to have equal rights from the start of the event. There is nothing more equal and fair than the pairings being randomly determined in the early rounds of an event (within a scoregroup).

The Swiss System is inherently slightly unfair. Over a short sample of tournaments it will likely be MORE fair than the results of randomized pairings, but over a long sample of tournaments the randomized pairings will be fairer to all players in the event (not just the top ones).

The issue is that chess players cannot see past the one individual tournament, and therefore they are happy to accept some small degree of inherent unfairness in order to assure that any one tournament isn’t too affected by “lucky pairings”.

Let me emphasize that once again I agree that the Swiss System is a fine system, and I’m not suggesting that it gets thrown into the trash heap. It’s completely normal that most Open events should continue to use it. But while I have your attention, let me try to improve upon the Swiss System:

Computers are pretty smart these days. It shouldn’t be difficult to set up a tournament so the first round or two is random, and then the computer could go out of it’s way to “equalize” the strength of opponents that people within a score group have faced throughout the tournament. So for example, if Kramnik and Caruana play in Round 1, then the computer will do what it can to slowly make their pairings easier during the remainder of the tournament. Something like this could be even more fair than the typical Swiss Pairings. Instead of using a system that ignores any luck of the draw from the previous rounds, this new pairing system would check to see who has had easier or harder pairings, and then alter the later pairings to make it more fair for those who had the tougher draw.

So imagine you have 4 players, two who have an average opponent’s rating of 2600 and two who have an average opponent’s rating of 2500, the computer would aim to give the ones who faced a 2500 average the higher rated opponents. If you’ve had easy pairings all tournament and are tied for the lead in the final few rounds, this system will try to give you the toughest pairings they can. Can you tell me any reason why something like this would not be more “fair” than the Swiss System?

What I’m suggesting is that when someone tries something different, and one which has obvious logical merits, that everyone doesn’t freak out because maybe one of the top players in the world got an unlucky break. In an open tournament everyone matters, not just the top 5 seeds. And by trying new things, we could experiment and perhaps find a way to improve upon the way things have been done for so long, such as my idea of a pairing system which specifically tries to equalize every player’s strength of opponent as the tournament progresses.

And lastly, the players all knew what the pairing system was and they agreed to play in the tournament. At the time of this writing, Caruana and Kramnik are playing an exciting chess game.

caruana

They are possibly taking more risk than they would have if the game took place at the end of the tournament. Why are we complaining about this!

9 thoughts on “Random Pairings Make Chess Players Go Crazy

  1. Greg – Great post! But some things should be clarified. What you’ve described occurring at Isle of Mann is indeed still within the definition of the Swiss System. What players have come to expect for many decades now are “Ratings-Controlled Swiss System” tournaments. With 9 rounds and 160 players, what Isle of Man is doing will work out just fine. Use of ratings for ranking make a swiss more efficient at weeding players down to an absolute leader most quickly, but it indeed creates many mismatches along the way. In the early 1980s, if I recall correctly, Bruce Diesen (mathematician, master, and cousin of IM Mark Diesen) introduced an alternative pairing system experimentally in Minnesota, in which players in the same score groups were paired with *closest* in rating, and players with low ratings and high scores were paired with players with higher ratings but lower scores. In round 1, pairings were 1-2, 4-3, 5-6, 8-7, and so on. Many more meaningful games resulted, but many more draws of course, too. A writeup in their state magazine described plenty of happy players, but the USCF executive director at the time railed against the system and I think made it more difficult to announce such events in Chess Life magazine – which in the days before the internet, was a killer. I thought it was a very promising system for relatively short amateur events.

    Like

    • I guess the knockout argument against all pairings random is that too many of the games will be between wildly unequal players, and not enough between players of roughly equal ranking. In an important public event, we do want to see Caruana play Kramnik sometime, rather than see each of them play some patzer.

      Like

    • @Jack Rudd Well said.

      I welcome innovation (and am constantly trying new things in the tournaments that I run) but it would be helpful if Greg first learnt about the existing pairing systems (such as Dubov, which is exactly what he described) and their known strengths and weaknesses. For example, it is well known that the Dubov system doesn’t work as well as the Dutch Swiss system when there are large rating discrepancies.

      I agree with Greg that the chess community is often too conservative and a spirit of innovation should be embraced but that doesn’t mean that all ideas are good ones. My view is that a better solution to the problem of often having a dull first round in big Open tournaments (due to the large rating differences) is to use Baku acceleration which gives players in the top half a bonus virtual point for rounds 1-3, which is reduced to a virtual half-point for rounds 4-5 before being removed entirely for rounds 6-9.

      Like

  2. Hi Greg,

    Allentown has used a version of “random pairings” for years — every few months, for variety, we run an event where the pairings are random. And when we say random — we mean random: you can play the same player more than once, you can get three colors in a row, etc.

    Now — we should clarify: our events are three rounds…and played at G/40 d5. And this is a club event, so while there may be wide rating gaps…if attendance is particularly good…we split the field into two sections and have (essentially) two random pairing events.

    The kicker — the prize fund only goes to players with perfect scores. Yep — you have to go 3-0 to get a share of the prize money in this format. The house roots against all 2-0 players in the final round.

    I agree that random pairings work best when there is not a wide range of ratings in the field. Our system occasionally allows a strong player to get a prize playing the bottom players. But then again, it occasionally allows the bottom players to get a prize by avoiding the stronger players. And it avoids the short draw syndrome completely.

    We like it — not all the time, but for variety. Greg is right about that.

    Like

  3. Pingback: Upset Alert: Tarjan Defeats Kramnik in Isle of Man - US Chess

  4. Pingback: Upset Alert: Tarjan Defeats Kramnik in Isle of Man - US Chess

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s