What is the proper action for a TD to take if pairing two players who have already played is unavoidable. If it is simply a case of lower than anticipated participation in the tournament, should the number of rounds be reduced before the start of the tournament? What if it could not be anticipated, due perhaps to mid-tournament withdrawals, or a spate of coincidental bye requests?
This is covered in Rule 27A1. “If it is necessary for players to play each other twice, then top priority should subsequently be given to having them face each other no more then twice.”
In one of the CCA Chicago Class events, the lowest section had six rounds and six players. The players were warned in advance that the final round would probably see players meeting opponents for the second time in the event (late entries might have avoided that).
In some of the scholastic class tournaments I’ve done the team trophies were determined at the division level (each grade-based division encompassing nine rating-based sections) while the individual trophies were at the section level. Reducing the number of rounds in a section would adversely impact their teams’ chances, so those players were also warned that they would end up replaying some opponents.
Having a long tournament may end up with it being more reasonable to repair opponents (at one of the extremes of the range of score groups) than to skip over multiple score groups to find previously unplayed opponents.
I don’t like the idea of reducing the number of rounds. For a four round tournament, if there are four players in a section I’ll have them play a three round round robin (more or less - see rule 29K) and in the last round it’s 1 vs 2, 3 vs. 4 rematches (1 being the top-ranked player and 4 being the bottom-ranked.)
If there are fewer than 4 players in a section in a 4 round tournament I’ll normally combine it with another section. In most cases where two sections are combined the prizes in the higher section become place prizes in the combined section and the prizes in the lower section become class prizes. Another option is that players can only win prizes from their original sections. I’ve done that in cases where trophies were engraved with specific grades, so only players in a particular grade could win the trophies for that grade.
Sometimes combining sections works and sometimes you get more (and somewhat justified) complaints that the lower section has first place determined by which higher section players the lower section leaders happen to be paired with.
Having the option to reduce the length of the tournament by a round and just do a round robin would make sense here IMHO.
Players meeting each other a second time is not really a Swiss, of course, but it isn’t a disaster. The players might find it a little unconventional, but so what? The “Australian Draw” system and various Ladder systems routinely feature players meeting more than once. As Ken Ballou said earlier, Rule 27A1 even allows you to call a section a Swiss and have players meet twice, provided the number of players is less than or equal to the number of rounds. In that situation your new paramount goal is to avoid having any players meeting three times!
Makes sense mathematically maybe, but not PR-wise. If players plunk down their entry fees for an N-round event, they expect to play N games.
If you cut the number of rounds, you have a truth-in-advertising problem. If you announce this possibility in pre-event publicity, you have a wishy-washy problem. Players who see stuff like this in the TLA might tend to stay away.
Bill Smythe
Cutting the number of rounds in advance also eliminates the opportunity to accept late entrants (unless you want to whipsaw back and forth by restoring a previously cut round), so it probably couldn’t be safely reduced by a round until after the start of the second round, in which case you have the first round played with different expectations (regarding the number of rounds) than later rounds.
It was paired as a round robin directed swiss, and meeting only one opponent a second time isn’t that bad.
A more drastically small section was a four section scholastic with about 150 players and only two entrants in the HS section. They ended up playing a four game match.
Changing the section to a round robin also runs into trouble if a player withdraws or has requested half point byes. (Of course, there are no half point byes in a round robin.)
Even worse is a four round section with three players. I’ve seen that in a high school section of a scholastic tournament. 
If there is a small, odd number of players (say 3, 5, or 7) and all players agree, you can have all players play two games simultaneously. For example, with 5 players:
Round 1:
A vs B
B vs C
C vs D
D vs E
E vs A
Round 2:
C vs A
A vs D
D vs B
B vs E
E vs C
This amounts to a 4-round, 5-player round robin without byes.
To compensate for the fact that each player is playing two games at once, you can double the time control.
Bill Smythe
My first thought was “Bill, you’re six days early.” I suppose it might work, though, if the players were willing to try it. How often have you done this, and have there been any problems with it?
I can see problems coming, of course. (“he had an easier opponent than I did and was able to spend more of his time on our game” claims)
With a small number of players, you can probably have the RR anyway (even though it will take 5 rounds), by doing ASAP rounds on a per-game basis (with breaks for meals as needed).
I’d never heard that expression, though I can guess its approximate meaning from context. 
When we had the club on Lunt Avenue, we’d have small 4-round events every Wednesday. You wouldn’t believe how often we’d end up with 5 players. So we did it a lot. I had the details figured out, including the table arrangement (see below). The players must have been OK with it, since they kept coming back.
Yes, I had thought of that possibility, but I guess the players simply accepted it as a matter of chance.
That’s certainly true, but it is a little slower, especially if you have one slow player in the group who has to wait a while for his round 2 game or something.
Instead of ASAP, I prefer STOP (Sooner Than Otherwise Possible).
If you want to try the double-round simul, the table arrangement is important. At our club we had 6-foot by 30-inch tables. We arranged five tables corner to corner, so that the 30-inch ends (important – NOT the 6-foot sides) formed a small regular pentagon in the middle, with the five tables sticking out in all directions, like the blades on a 5-blade ceiling fan. Each player would sit in one of the V’s between the “blades”, playing white against the player to his left and black against the player to his right. (Nobody sat inside the small center pentagon, of course.) All clocks faced toward the center, so that all players could see all clocks.
Of course, there is only one game per 6-foot table, so you could just as easily use five square tables instead (e.g. card tables 30 inches on a side). For that matter, you could mix card tables and long tables, as long as you have that small pentagon in the center.
In the first double round, pairings can be random. In the second double round, simply seat each player adjacent to each of the two players he has not already played. (With 5 players, it is a theorem that this will always work.)
Bill Smythe
If you will allow me a regionalism, this is just wicked cool.
My old chess club did this periodically by design, although normally we did it in groups of three.
We called it “Concurrent Chess”.