…
Going too fast usually gets you in trouble!
Sometimes I’ve seen Swiss-Sys make some insanely stupid pairings that are just begging to be changed. A couple of months ago I was playing in a four round tournament. In the first round there were no upsets except a 1720 drawing with a 2180. There were even numbers in all 3 score groups. Swiss-Sys paired the 1 pointers normally. The two 1/2 pointers had played each other so the 2180 got paired against the highest 0 (1970), or perhaps the 2nd highest 0 (1950) to balance colors. I’m not 100% sure about the color issue. Did the 1720 get paired against the next highest 0? NO! The 1720 got paired against a 1739 who was on the bottom half of the zero score group. The computer bypassed the entire top half of the zero score group to come up with that pairing. There was absolutely no logic to the pairing since there were plenty of higher rated players in the score group who were due the correct color to play the 1720. Since the tournament was running behind schedule the TD had not checked the pairings the computer made. Normally in the 2nd round of a 30 player Swiss one would not expect such a dumb pairing to crop up.
Both Swiss-Sys and Win-TD have been known to have bugs that cause oddities like that. Usually they occur in a new version, and get fixed once the programmer has been alerted to it. I think one needs to at least make a cursory review of the pairings to look for the obvious error like my example. In later rounds when more is at stake, one must be careful, and spend time looking at the pairing logic before making any hasty decisions regarding switches.
Today, this is good advice. 5 years ago it was not. A few recent incidents got me to thinking, and I must say that I have not seen a pairing done by WinTD in the last 5 years that needed to be changed.
Alas, I (very) recently saw some pairings changed, and the logic used by the TD who changed them was very suspect. To my mind, the logic this TD gave was the best evidence I can think of for NOT changing competent computer pairings. In fact, the TD took pairings that were perfectly correct and changed them so that at least one black-letter law of pairing was violated. But, it was a law broken in a way that the TD was confident would not lead to a complaint.
When pairings get difficult, it is generally impossible to come up with the correct pairing by looking at only a few players, or a single scoregroup.
The good news is that the proliferation of competent pairing programs means that fewer and fewer people can do pairings manually (either with cards or simply studying the wallchart). That applies to TDs and players. The good news is that the TDs have computers, and the players don’t. So…complaints about pairings are generally decreasing.
But, we still have a few TDs who make “defensive” pairing changes - intended NOT to improve the pairings, but instead designed to avoid complaints. The downside of this is that these defensive pairings often take an imagined downside for ONE player and spread the pain around by penalizing MANY players (in the hope that each individual harmed player won’t notice). That’s seen as easier than standing up to the ONE player who is highly likely to complain.
Today, it seems to me that the best rule is: computer pairings stand unless there is an OBVIOUS inequity that MUST be rectified. In no circumstance should a TD make a change because another pairing seems PREFERABLE.
I’ll get in trouble for this suggestion - but let me put forward the idea that being able to pair difficult situations by hand is no longer a prime TD skill. Of course TDs should understand the (current!!!) pairing principles and be able to explain them - but the ability to pair by hand (again, difficult, tricky situations) is now more useful for showing off in the back room than it is in running a successful tournament.
As recently as 5 days ago I played in a tournament where the TD had to correct the pairings produced by WinTD. In round 2 of a 2 round tournament paired by the Schroeder method, WinTD did not pair 1vs2 in any of the score groups. The Local TD appropriately recognized the mistake and corrected the pairings by hand.
I have corrected other WinTD pairings in the last 5 years, although rarely. The usual scenario is a small section with an unusual situation that seems to throw the software into a loop.
One of the most frustrating chess moments for me was at a small tournament about 2 hours from my home. In the first round the TD using SwissSys paired me against one of my students that I had brought with me. I asked that the pairing be changed and the TD attempted to comply. When he realized that he didn’t know how to switch a pairing he stated that he was going to let the pairings stand. My offer to help perform the switch was refused.
Just remember Clarke’s Third Law:
Then there’s Gehm’s Corollary to Clarke’s Third Law,
I do have to wonder in the case of TDs if we aren’t building a generation of TDs who are what H. G. Wells referred to as Eloi, people who use the machines but have no concept as to what they are or how they work. I guess (or maybe hope) that makes me a Morlock.
Bill, Ken or Grant, Please give some info on the Schroeder method, its author and uses.
Page 170 in the rulebook–1 Vs 2 pairings is a good start. We used this at the club a lot when we still had a meeting sight.
Is it possible for you to send me the crosstable for this event? Failing that, please send the name & date of the event and I’ll find it on MSA. I have a few theories about what happened, but would like to double check the crosstable to be sure.
I believe it was named after James Schroeder, formerly of Ohio currently Oregon.
I use the method a lot in non-rated scholastic sections. We assign local ratings based on grade and then pair 1 vs 2. Thus 1st graders will play 1st graders in their section, 2nd graders will play 2nd graders, etc. This way we don’t end up with a 1st grader playing 3rd graders the whole tournament.
We just have a K-1 and a K-3 section to accomplish that. 
I am usually the TD and not organizer, so I don’t always have control over the sections.
I have been told by an educator that developmentally its the K-2 players who should be separated from the 3rd graders, so perhaps you should consider a K-2 section as well.
You might even use accelerated pairings. 1 vs. 2 is kind of meaningless if the original rankings are as coarse as “grade 1”, “grade 2”, etc.
Accelerated pairings are most often used to get a clear winner from a “too big” section. They can also be used to avoid gross mis-matches in the kind of K-3 section Grant has.
On the third hand - if all you know is grade, and see enough good 1st graders who keep beating 3rd graders though the entire tournament, I might consider RANDOM rankings and ordinary Swiss pairings.
I’m not a big fan of 1-2 pairings, esp. if you follow all of the other rules on color balance and alternation. It always seems to me that you lose all the benefits of rating information (or, worse, do the WRONG think with the rating information) and you don’t simplify things. I actually prefer “color blind” 1-2 and might even prefer “results blind” 1-2 (1 & 2 keep playing as long as they keep drawing).
With a non-rated K-3 section, rating information is next to nonexistent. By ignoring any “real” rating information and assigning ratings based on grade, e.g. 1st grader 101, 2nd grader 102 etc, 1vs2 will pair the 101’s in a score group against the other 101’s, the 102’s against 102’s etc. In the later rounds the “better” 1st graders will be paired against the better 1st graders and ultimately against the better 2nd graders. I rarely see a 1st grader playing against a 3rd grader, they usually lose to the 2nd graders.
A potential argument against this method would be the increased chance that a player ends up with a trophy that wouldn’t have otherwise. The upside is a better experience for the younger players. I have used this method for the last two years or so, and have had many parents comment that they were appreciative of the fact that their child didn’t have to play older players.
Right - but you get the same effect by “accelerating” the pairings (and lessen the probability that you will screw up the section parameters given to WinTD, or expose a problem with WinTD’s pairings). The idea is to use a method which give the desired effect AND has been well tested.
While I can live with the increased likelihood that we will have more perfect scores at the end of the tournament with accelerated pairings, I don’t like the overall pairings that result. With accelerated pairings in the second round the first round winners will be paired against first round losers. I would rather have the kiddies who don’t know how the horsey moves to be playing against each other as early and as far away from the rated kids as possible.
But…you will have the experienced 1st graders playing against the beginner 3rd graders. If you are going to have any 1st-3rd grade matchups, those are the ones you want. If you don’t want any 1st-3rd grade matchups, then you should (in my opinion) have separate sections.
You either believe that 3rd graders should be treated as higher rated than 1st graders…or you don’t. And if you do, then I don’t particularly like making the stronger players play each other early in the tournament.
Acceleration seems to me to be a good compromise: in the first round you pair the older players as a group and the younger players as a group. Once you have a little bit of information you can pair the bottom-performing older players against the top-performing younger players. These are much less likely to be walkovers than top vs. bottom in the first round - but not so close as to be inappropriate pairings so early in the event. By the 3rd round, you have separated the sheep from the goats, and the very strong and the very weak players are well isolated from each other. At the same time, you increase the chances that the tournament-winning decisive game appears in the last round rather than the first round.
using 1-2 pairings, the strong 1st grader is given a free ride to the later rounds.
But…I’m still confused about one point. Are you using the 1-2 pairings described in the USCF rulebook…or are there further variations (color, previous results, etc.)? Perhaps that’s why WinTD did something you didn’t expect in R2?
Has anybody seriously considered Verber pairings? It seems as though it should get rid of a lot of perfect scores quickly.
In a large (for its day) scholastic tournament in Chicago in the 1970s, with straight pairings, there were still about 30 perfect scores with two rounds to go in the 5-round event. Richard Verber took over the pairings for round 4, and extended the idea of “lowest 3.0 vs highest 2.5” to its extreme. He paired the bottom whole-bunch-of 3.0s against higher-rated 2.5s, 2.0s, and maybe even 1.5s. It worked perfectly – after round 4, there were only three perfect scores. After round 5, there were none – two of the three drew each other, while the third lost to a higher-rated lower-scoring opponent.
These seat-of-the-pants pairings could be codified along the following lines:
Round 1: 1-vs-2 pairings. These will produce a lot of higher-rated losers and lower-rated winners.
Round 2:
2A: Winners within 300 rating points of the top player are paired against each other, 1-vs-2 style.
2B: From there on down, winners play losers, one by one. For example, the top winner not already paired in 2A is paired against the top loser, etc. The result will be higher-scoring lower-rated players paired against lower-scoring higher-rated opponents. With about a 300-point rating difference all the way down the line, most of the winners will lose, and most of the losers will win.
2C: The remaining players (losers) are paired against each other, 1-vs-2 style.
Rounds 3 and following: Continue the method begun in round 2.
I’d bet this would wipe out perfect scores extremely quickly – much faster than either normal or accelerated pairings.
Of course, some details such as what to do about draws, how to handle odd players, colors, etc would have to be worked out. But I’ll leave that as an exercise to the reader.
Bill Smythe
Why should the primary goal of pairing be to reduce the number of perfect scores?
IMHO: So that when non-money indivisible prizes are awarded tie breaks will not have to be used. At scholastics it is a hard sell if there is a three way tie for first and you have to explain to the two tie-break losers and their parents/coaches why they did not get the trophy. In their minds they scored just as well as the trophy winner and the tie-break system is just some magic that cheated them out of a prize. It works out a lot better when you don’t have to use tie-breaks.