John, this was a potentially interesting and refreshing post for me since I’m no great fan of swiss pairings (but I don’t advocate their downfall either). When you say “I think it still is,” are you saying you the Holland System was the best that could be done outside of the Swiss System - or do you even mean including the swiss system? If the latter, I’ve never heard anyone say that before and would love to hear your take on the matter. If the former, well, this will be a short thread, huh?
No, I meant that the Holland System was the best way they could find to deal with large groups before the introduction of the Swiss. It still has some specialized uses (see previous thread), but the Swiss replaced it because it worked better.
Is that the same as what is called the Dutch system by FIDE? Someone must still be using it, when I was at the FIDE Congress in 2006 they were debating changes to it, as I remember it was something to do with how to handle the odd man, possibly when there are unplayed games involved for several people in the score group…
I suppose you could say that a Holland is better than a Swiss, if the Holland has 8 rounds and the Swiss 5. Given any N, however, an N-round Swiss will do the job better than an N-round Holland.
For those readers who are unfamiliar with a tournament format that is little used.
A Holland System tournament is a multiple stage round-robin event (similar the the soccer World-Cup)
Whare the players are divided into multiple Round-Robin sections then the winners (or a predetermined number of qualifiers) advance from each section to a Round-Robin final section. Depending on the size of the event, two preliminary stages may be needed.
As stated, this can handle many more players than a single round-robin, but still requires many more rounds than an equivalent number of players in a Swiss event.
I believe that is one variant of pairing a score group. It is the one recommended by Kenneth Harkness when he advocated use of the Swiss in the late 1940s or early 1950s. It had to wait until there was an acknowledged rating system, since the ratings control the order of the score group. Other methods include top vs bottom (i.e., in a 10-player score group, 1 plays 10, 2 plays 9, etc. with 5 against 6 for the final pairing), or 1 vs 2 (1 plays 2, 3 plays 4, etc., up to 9 playing 10). Those are options listed for FIDE-rated Swisses listed in a British “Handbook for Organisers”. Other FIDE variations in Swiss pairings include order of pairing score groups (USCF rule is highest to lowest, FIDE order is highest, then lowest, then next highest, then next lowest with the even score group paired last. The other major FIDE variant is handling of the odd man (float in FIDEese). FIDE rules state that in groups with greater than 50% score, the odd man is the lowest-rated and moves down to the next score group (the same rule for USCF pairing rules) and in groups with less than 50% score, the odd man is the highest-rated and moves up to the next score group.
What does FIDE say about who the player is to be paired against when he is floated down? I can think of at least four reasonable answers:
He is paired against the highest-rated in the next lower score group.
He is placed at the top of the next lower score group, thus causing him to be paired against the middle player in that group.
He is inserted into the next lower score group in the spot dictated by his rating.
Same as 3, but his rating is adjusted (say by 100 points) to compensate for his extra half point.
I should point out that if answer 1 is used, it makes no difference whether you are floating the higher-scoring player down or the lower-scoring player up.
This really won’t make much difference in the end. Score groups near the top and bottom tend to be smaller, so there are likely to be more “rematch” problems. The FIDE method seems to address this possiblity directly. In practice, though, simply going from top group to bottom group works fine in practice, although occasionally you may have to “unpair” a score group and do it over because of problems in the next lower group.
Bill, your first assumption is correct. For an up floater, the highest rated in the lower score group plays the lowest rated in the next group up; the opposite of the normal lowest-plays-highest-rated for greater than 50% groups as well as normal USCF odd-man procedures. For the greater than 50% groups, the floater is treated exactly as in USCF odd-man pairings: the lowest rated in the higher group plays the highest rated in the next group down. I was in a rush to post my reply so I didn’t relate all info regarding FIDE vs USCF differences in Swiss pairings.
In that case, downfloating the lowest player is always equivalent to upfloating the highest player. This is true no matter whether plus or minus score groups are being paired.
I must, therefore, conclude that FIDE’s suggestion (to pair the highest and lowest groups first, and the middle groups last) is much ado about nothing. You get the same results either way – they are not “opposite” at all.
