WinTD has an option to use decelerated swiss pairings, where in round 2, round 1 winners face round 1 losers. They suggest it as an option when you have about one extra round for the number of players, for example 8 players in a 4 round tournament.
Does anyone have any experience using this system and advice about whether or when to use it?
Grant Neilley
I have never tried “decelerated pairings”, but I do know that special considerations come into play in pairing small tournaments.
For example, in a 6-player tournament with 4 or 5 rounds, if all the colors alternate in round 2, and again in round 3, then there are no possible pairings at all in round 4 (without pairing some of the same players twice).
And in an 8-player tournament, the same is almost true. There will be only one set of (grossly unnatural) pairings.
Bill Smythe
My concern is whether decelerated pairings would increase the probability of really BAD pairings in the late round, or no valid pairings at all.
That’s the sort of thing that you could try testing, running a number of test events.
Is that the system where the top quarter is paired against the 2nd quarter with the 3rd against the 4th? In that case the top half non winners are paired against the bottom half non losers for the 2nd round.
That’s accelerated. Decelerated is something I’ve never tried.
I think the idea of decelerated pairings is that they would decrease these disastrous possibilities. Of course, decelerated pairings should be considered only in a section with way fewer than the theroetical 2^n players (n = number of rounds).
Bill Smythe
Is there any situation where you would prefer decelerated pairings over a RR-guided Swiss?
When it is an individual/team event. The RR-guided Swiss would not avoid pairing teammates.
Probably not, personally. Maybe Tom Doan could jump in here and defend his idea.
Bill Smythe
With a very small number of players, not pairing teammates is #1 on my list of compromises to make.
Note that you can still avoid pairing teammates ON THE TOP BOARD using a RR-guided Swiss (at least, until you are eventually forced to pair teammates there, as well.
We we discussing events where the small number of players sometimes leads to pairings traps. Trying to avoid pairing teammates just makes the effective number of players smaller - and makes those traps more likely.
The original post was not about small tournaments, that was a tangent taken afterwards.
I have never even heard of decelerated pairings before. My speculation is that they might increase norm opportunities when you have a fairly even split of top half titled, bottom half untitled players. Even going (0,1) in such a field, the untitled player would then get a third titled player in a row.
I think the term “decelerated pairings” was invented by Tom Doan, as a means of handling very small tournaments. He liked his own idea so well, he included it in the next version of WinTD.
Bill Smythe
So much for my speculation. Now I get your Tom Doan reference from before, and the small tournaments wasn’t a tangent at all.
I guess the question is how well it works. Because a controlled round-robin definitely works.
With a large number of rounds for the players, but still not approaching a round-robin (say 12 rounds for 30 players), I always paired down to the median score group. Then I started from the bottom, giving the lower ranked player his due color, and worked up to the median score group. Finally I paired the median score group normally. This is 29I1 with a twist. I also used this system for club pairings (say 4 rounds for 20 players but with lots of byes) up until switching to Swiss-Sys. But I don’t think this method would help much at all with 4 rounds for 6 players.
I liked your explanation here: http://newsgroups.derkeiler.com/Archive/Rec/rec.games.chess.misc/2006-08/msg00462.html.
This is neither accelerated or decelerated, it’s just another way of arriving at standard pairings (except that a few colors will be reversed). It has the advantage that you may avoid having to re-do pairings in (for example) the 1.0 score group because the only two players in the 0.5 score group have already played each other.
I’m not even sure what decelerated pairings are, perhaps Tom Doan explained it in an old thread somewhere in this forum. Maybe it works something like this:
When pairing round 2, ignore the scores after round 1. In effect, play another round 1 (transpose slightly so that players don’t face the same opponents again). Similarly, when pairing round 3, use the scores after round 1, but ignore the round 2 scores. In general, when pairing round N, use the scores after round N-2 instead of those after round N-1. (Do, however, take colors in the “missing” rounds into account, and don’t pair the same players twice.) Eventually (say, in the last 2 rounds), stop ignoring the previous round and make normal pairings.
So, perhaps I should change my earlier answer to Prof. Sloan:
It could be that decelerated pairings might be the best way to handle your example of 12 rounds, 30 players. Or, say, 7 rounds, 20 players.
Bill Smythe
If that’s decelerated pairings, then it’s very similar to the way I used to pair blitz tournaments. I called them warp-speed pairings, and the idea is that you pair the next round while the current round is under way.
Round 1. Pair 1/4 against 2/4, and 3/4 against 4/4.
Round 2. Pair 1/2 against 2/2, without considering round 1 scores.
In other words you reverse the theoretical first and second round pairings.
Round 3. Pair using scores from round 1, colors and opponents from rounds 1-2.
Round 4. Pair using scores from rounds 1-2, colors and opponents from rounds 1-3.
etc.
The theory is that in a blitz tournament the worst part is waiting between rounds. This method keeps things moving right along. So at the cost of slightly worse pairings, the players get much faster pairings. It’s not necessary to play the final rounds in the “correct” order because of the randomized blitz results. (A variation is to wait for the penultimate round to finish before pairing the last round. I did it that way at first but it turned out not to help in practice.) With enough rounds the top players all play each other anyway. Of course with a blitz tournament it’s not a problem getting enough rounds.
By the way, for a blitz round-robin tournament you NEVER let the players choose the next opponent based on who finished first. That chaotic method lengthens the tournament by about a third, as measured on the wall clock. What happens is at the end you have a lot of players standing around all waiting to play the same opponent – usually some idiot who disappeared in the middle for a long bathroom break.
The right way is to designate a “robin”. After all the games in a round have finished all the players, except “robin”, move over one spot. There are a few more details … if anybody doesn’t already know how it works I can explain. The hard part is convincing the players before they have used this method that it’s better for them. My solution to that is basically take it or leave it. Once they have seen it in action, though, they think it’s great.
The disadvantage to both of these methods (swiss and RR) is the TD can’t play.
With computers, you don’t even have to wait between rounds any more. I ran blitzes with 20-30 players and there is only 20-30 seconds between rounds. The only seeming advantage to your concept is that 2nd rounds games can start before all the 1st round games are over, but that leaves them players waiting longer later as it builds like that.
And with the computer, the TD can play in the blitz (o;
When I learned to drive, there was only one kind of automobile license (class D in Massachusetts). Later they instituted an automatic-transmission-only restriction. To get the unrestricted license you had to take your test on a manual-transmission car.
Are we approaching the time when the USCF should have a computer-assisted restriction for TDs?
Absolutely not!
A TD who tries to run an event using a computer when he does not know how to run the event withOUT the computer…should have his license lifted (or, never granted in the first place).
Computer assistance is IN ADDITION to the normal TD skill-set.
It would make more sense to have a “no computer” TD level - and another exam to qualify for computer use.
In other words - the computer is a motor-cycle. First you need a regular license, AND THEN you need a license for the advanced stuff.
TDs using computers are addressed in the TD Certification rules. Are you suggesting a change?