USCF Rule 28R says that, “where the number of players far exceeds the number two raised to the power of the number of rounds… The director has pairing options…” and goes on to describe accelerated pairings.
The phrase “far above” is subjective, obviously, but let’s suppose there is a five round tournament that has fewer than 32 players… in this case, can accelerated pairings be used anyway?
Or, as the rule implies, are accelerated pairings only an option when the number of players of greater than 2^Number_Of_Rounds?
Some players (and TDs) don’t seem to understand accelerated pairings. They just assume “acceleratedpairings” (thought of as a single word) is a magic panacea to somehow make pairings “better”.
For example, I have heard it suggested that pairings should be accelerated “only in round 1, not in round 2”. Or that “the acceleration should be continued in round 3”. Nobody who makes such suggestions seems able to actually define how they would pair round 2 (in the former case) or round 3 (in the latter case).
Accelerated pairings accomplish one thing only. They reduce the chances of having multiple perfect scores at the end of the tournament. They do not reduce the chances of having multiple winners. In fact, they may increase these chances, with nobody finishing with a perfect score and a slew of players with 0.5 less than a perfect score.
Accelerated pairings also have a huge negative. For the vast 75% majority with 1 point after 2 rounds, round 3 feels pretty much like round 1 would have felt had the pairings not been accelerated. This just feels yucky.
In some non-USCF tournaments I’ve done a partial acceleration in round three after doing sextile acceleration in rounds one and two. Most recently it was done in a fixed-weighted-board team tournament with 44 teams, four rounds and a very small chance any match would end in a draw (the eight boards were weighted to be worth 68 points, 12 pts to 5 pts varying by one per board). Without acceleration I would expect 11 teams at 2-0 finishing with three teams at 4-0. With sextile acceleration I had hopes of only four 2-0 teams but after two rounds I had six teams at 2-0 with two of them winning round two matches that they were the underdogs in. I paired those last two teams against the top 1-0 teams (they lost) and thus entered final round four with only two 3-0 teams and finished with only one 4-0 team.
PS For players in the middle acceleration can often result in more mismatches than not accelerating, due to the wider middle.
My original question concentrated on whether you are allowed to accelerate pairings, as per rule 28R. This discussion has already veered into the value and reasons for pairings.
So let me ask a different question. Is there a rules committee / governance structure that can answer the question?
I believe it’s entirely at the TD’s discretion—I’ve certainly seen it applied when the number of players is below (not just way above) 2^rounds. I would agree with Alex that that probably should be seen as a major variation, particularly in a money tournament where the prizes are divisible anyway. (Round three pairings for the 2-0’s can be wildly different in quality depending upon how many bottom half players win their first two. In a trophy tournament, that tends to work to the benefit of the strong players who play tough round three opponents and will benefit in their tie breaks. In a money tournament, it works to the benefit of the 2-0’s who get easy round threes since a point’s a point.)
As I suggested above, should a TD/organizer try this without advance notice, that is what special referees are for. If there is advance notice, he can do just about anything he wants.
As I’ve suggested several times previously, this is the one big advantage of playing by US Chess rules. In nonrated tournaments, presumably the decision of the TD is final. In rated tournaments you can appeal bad or unfair decisions, all the way to the Delegates if necessary.
Given this, the “conditional” wording sounds like a rationale not a condition. Thus, one might argue that accelerated pairings might be available for players less than 2^n # of rounds. Probably, this should be clarified, and unless the number of players exceeds 2^n shouldn’t be used - or if we want a buffer, X% of 2^n. Perhaps X% is 90%.
Also, the rationale says “may not” and this could be clearer with “might not”. We don’t wish to imply that permission is needed, its simply pointing out that it possibly won’t occur. It’s also not clear to me that the “rated” in red above shouldn’t be “ranked” (since the odds of winning are purely dependent upon rating, not upon ranking.)
If a five-round tournament has 24 players then without accelerating the maximums are: twelve at 1-0; six at 2-0; 3 at 3-0; 2 at 4-0; 1 at 5-0. Quartile accelerating makes the maximums: six upper half at 1-0 and six lower half at 1-0; three upper half at 2-0 and six lower half at 2-0; 5 at 3-0; 3 at 4-0; 2 at 5-0 and sextile accelerating would vary by having a maximum of ten at 2-0 instead of nine. Always remember that if acceleration doesn’t work well then it can increase the number of 2-0 players by 50% (versus the 50% reduction when it works very well).
Quartile acceleration of 24 players that works well (higher rated always wins) can have twelve at 1-0, three at 2-0, two at 3-0 and one at 4-0. Sextile could instead have two at 2-0 and one at 3-0.
Even if you have “enough” players to justify acceleration, I’m inclined to think it’s not a good idea if the ratings of the players aren’t deemed reasonably reliable. For instance, in an under 300 section, you’re liable to make things worse because you may not get those round two wins from the second quartile over the third quartile. Similarly in a class tournament, if most of the players are within 200 points of each other you’ll probably make life worse.
For accelerated pairings to work you need a large spread in the ratings, and you need the ratings to accurately reflect the players’ skills. They are contraindicated in cases like a class tournament where most players ratings in a section are relatively close to each other, and also in cases where you have many provisionally rated or unrated players. As Mr. Wiewel has pointed out upthread, if you use accelerated pairings in a situation where they don’t belong you likely will end up with more perfect scores than if you had just paired normally.
Bill Goichberg once told me of an experiment he tried in a high-school event where almost all players were unrated. He couldn’t use accelerated pairings directly, with so many unrated, so instead he tried pairing 12th-grade losers vs 9th-grade winners in round 2. It backfired badly. The 9th-grade winners crushed the 12th-grade losers.
Players seem to believe that the purpose of accelerated pairings to avoid “unfairness” of a swiss. It does delay it to round 3. The sole purpose of it is to avoid multiple winners in large fields. Use when needed.