Maybe this belongs in the article section, but since it is pairings I am putting it here. The article talks about an alternate pairing system for a small tournament. Random pairings to be used to avoid massive rating differences. The article fails to talk about the class pairing option. In the final round players are paired by point group, but also within the same rating class. I do not remember the last time I saw this option used. Also, what about those players that prefer to play a stronger opponent? Doesn’t the random pairing option take away from those players?
The article didn’t mention class pairings because that is an entirely different subject. As to the players who want stronger competition, well, they can decide on their own whether random pairings undermines their desire. That’s why you have to announce it in advance.
I remember somebody trying this type of pairing system in the '70’s. As I recall, a lucky Class C player went 4-0 and avoided playing any of the Experts in the event. In the last round he beat another lucky Class C with a similar score, while the Experts, who had a harder pairing schedule throughout the event, drew their game. Since there was only a first prize and no other place prizes and few class prizes, there were several disgruntled players at the end of the event. I did not see this system tried after that. This one result should not be enough to avoid using this pairing system or one with perhaps two rounds of random pairings for a small event. As the author noted, when there is a small rating pool, it is possible to play the same group of people all of the time. It is pretty bad when you play the same guy in the first round every tournament.
The bigger problem is the possibility of color problems in the last round of a 4 round event. You can have a small score group with “lucky” players who both have WBB. After reading the article, I did a few simulations of this using pairing cards for 16 to 20 players and had it happen frequently enough that it is a problem. Breaking up their pairing affects the true randomness of the other pairings, and may not solve the problem anyway. You may have to ignore the rule giving the same color three times in a row, or just let the players toss for color, which seems a fair way to resolve the problem.
The few times I saw class pairings used they were not used for any player who was in mathematical contention for a place prize. The C’s should have faced the experts.
As I recall there have also been tournaments that used ‘ladder’ pairings. 1-2, 3-4, 5-6, etc. for the first round of an event. I would think for a club event that this might be preferred to random pairings. You totally avoid the big rating difference that a swiss gives you, and that can still occur with random pairings; at least for the first round. I do not know the exact details of such an event, and how long you retain the ladder-like pairings.
I do recall that it once was common to do class pairings in the last round to determine prizes rather than follow the Swiss System rules to the letter. If the pairing did not affect place prizes, then players in the class would fight it out. This led to fewer mismatches and tougher pairings for the top players, which they did not mind.
I am used to Swiss tournaments where each USCF Class (or Elo range) is its own separate tournament.
So I assumed the article was talking about switching to Random pairings, but still within Class.
---- (EDIT: After seeing MRegan’s feedback, and rereading the article, I see this portion of my post is wrong. But I am leaving it rather than confusing people by deleting it.)
Basically the Swiss system appeals to Socialist culture, while the Random system appeals to the Individualist culture. (Something like that.)
Swiss = “To each according to his need.” And a player at 0-2 in the standings needs a weaker opponent in round 3 in order for the player to enjoy the tournament better.
Otherwise, Random pairings with a 1000 Elo range would work only if the number of rounds was very high (far beyond the point of practicality).
More fair than either Swiss or Random might be a ‘Median Elo’ system.
In the USCF Class B section of a large tournament (Elo 1600-1799), the system could tend toward a final outcome of all pairings wherein every Class B participant would have about a 1700 average for the Elo’s of all his opponents.
A bigger problem that causes unfairness is the overuse of USCF Classes A B C D for the selection of Elo sections for the tournament.
Because in large tournaments the USCF Classes are used for section definition almost exclusively, all 1650 players are thereby arbitrarily placed at a permanent disadvantage compared to all 1750 players, in terms of prize winning.
Elo ratings are often stable enough to make this a real problem.
More tournaments should have Elo sections of:
U1900 , U1700 , U1500 , U1300
…instead of almost always using…
U2000, U1800 , U1600 , U1400
Gene,
The idea is random within a score group. Not random within the section.
The tournaments we organize alternate between even and odd 100 point cutoffs. CCA alternates too. So, I don’t understand your claim that even 100 point cutoffs are used “almost exclusively”
Mike Regan
But I am leaving it rather than confusing people by deleting it.)