Maybe I should clarify a bit what TDX is - it is a simple, freeware tournament management app I recently released, written in Java so it will run anywhere. As for pairings, it follows these simple rules:
Players can’t play each other more than once
Players play those with the same score unless:
A player can’t have the same color 3 times in a row or one color 3 times more than the other
That’s pretty much it. Odd players in a score group (or a player that can’t be paired by the above rules in a score group) get moved down to be paired against the highest rated player in the next score group for whom there is a valid pairing by the above rules.
Tonight I added a flag to implement 29E5f1 variation, and I’m sure I’ll end up adding a lot of similar flags as TDX continues development.
My own thoughts on this agree with some of the comments I’ve seen up to this point - 29E5f1 is a variation, and I think that the decision TDX made was valid if that variation is not being used, as odd as it seems to not have the only two undefeated players play each other. Of course, as a complete novice TD, I could be wrong.
Later: I’ve been giving more thought to this, because the business about equalization is a bit confusing.
Neither 27A nor 29E5f mean that you can break a scoregroup for equalization. What they are saying is a) you can’t break a scoregroup just to avoid three in a row, and b) you can give three in a row if it is necessary to equalize colors. The latter case can only come up in a very long tournament. For example, suppose you had to pair two players whose colors were BBWBBWW and WBWBWBW. (We are assuming that you have already tried all possible transpositions.) The first player has had four Blacks and three Whites, the second four Whites and three Blacks. So the first player has to get an equalizing White, even though this means giving him three Whties in a row.
Variation 29E5f1 is a compromise with the 3rd-edition rule. It says that you may break a scoregroup to avoid three in a row except in the last round. So, if this variation were in effect, if, in round 5 of a six round event you had two players tied for the lead with colors BWBB, you would not pair them (because three in a row is forbidden), but would instead drop them to the top players in the next scoregroup. However, if it were the fifth and final round, they would have to be paired, with one getting three Blacks in a row. The bottom line is that the leaders have to play in the last round if possible. This is fundamental to the philosophy of the swiss system.
Another way to put it (and probably the way a computer would implement it) is this: The hierarchy of rules to follow in making pairings is 1) haven’t played, 2) same scoregroup, 2) equalize colors, 4) avoid three in a row, 5) alternate colors. (After that it gets more complicated.)
Although the introduction to 27A implies otherwise, I don’t know of any published variation which gives equalization higher priority than pairing within the scoregroup. (Though of course you could design one if you wanted to.) Aside from 29E5f1, which we’ve already covered, 27A cites variation 29E5H, which gives equalization priority over rating. In simple terms, this just means that the 80- and 200-point rules are ignored, and you equalize as many colors as mathematically possible.
Rule 29E5f1 is a good rule. It was designed for directors with small entries and a large number of rounds. Did have a 5 round swiss with 9 entries, the 5th round pairings were not that nice.
If I did the pairings by hand, I would have flip the players colors around during the first number of rounds. If you had white in round one you should get black in round two. The ideal color history would be WBWBW or BWBWB with 5 rounds. When looking at a 4 - 0 player with the final color history of BWBBB, it would not have been hard to change the colors in round 3 or round 4. If I can look at a players ratings, can give a sniff test to see if the player has a better change to win the event. Since the 3rd round has a great amount of players, could change the pairings around to give the top player the white. If that did happen, the final score group would have been BWWBB.
Do not see a big problem with the software, it could be missing a key input of the ranking of the players. As it looks like the software did not have any data of the players rating. Its’ still possible to have someone have a bad color history. Since you had 70 players you had 35 boards, I know its’ going to sound harsh: its’ better to have someone on board 35 have BWBBB then board 1 have BWBBB. Since the top boards in the last round are the trophies and prize awards, would have more bitter and upset group of people. If the player that had BWBBB did lose the final game, everyone in the tournament and supports for black will question not just the pairings of the game but all the pairings. If the bad color history was given to someone on board 35 of BWBBB. Since the players on board 35 was out of the prize, it would have taken out the effect of the prize out of the matrix. Look, no director wants to give 3 blacks to a person.
29E5f1, whether being used or not, does not apply in the final round. It exists as an option to break up the score groups for all other rounds, but not the last round. The only way you can’t have two clear leaders playing each other in the last round is if they have already played each other. Period.
After reading over all the discussion on this, the light bulb has finally turned on over my head. I’ve adjusted TDX accordingly to: Pair players in score group by finding first player not yet played against and that doesn’t violate having a color 3 times more than another color or 3 times in a row. If that is not possible, then pair against first player that wouldn’t result in a player having the same color 3 times in a row. If that is not possible, then just pair against first player possible. I’ve also added a flag to allow score group splits to prevent 3 colors in a row or one color 3 times more than another, except for the last round. I think that covers the basics.
Now, what about teams? My team flag is currently just like the already played flag - it can never be violated.
Alex is right to have the pairing of the director override the computer software. Both the pairings, computer software and the directors pairings are not ideal. Only seeing the error that lead up to the final pairings for board 1, was the random roster. There is a clear reason to have the ratings in the pairings.
If the software did not get the data of how stronge the players were in the first place, it would change the score group. Think of it this way, say you have a master. If the roster is random, the master could have been paired up with masters in the first two games – then lose them both. That would place the master in the score group of 0.0 after two games. The other person can be a class C player, the first two games could have been paired with a class J player – then win both games. So in round three the class C player is 2.0 and the master is 0.0.
Having a random order, does not mean you would be paired with the same level of player as yourself in the final round. With it being random, did make the pairings off for all the games.
If that’s what you want, fine. Note, however, that at some point you could easily end up with no legal pairing this way. (SwissSys gives you a query for this, and if you say “continue” it will pair some players twice, following the other pairing rules as far as possible.)
It clearly DOES apply. It’s the only situation where it would apply.
Your last paragraph is correct.
“Variation 29E5f1. Last-round exception. Except for the last round, when it may be necessary to pair the tournament or class leaders, players shall not be assigned the same color in three successive rounds”
No, 29E5f1 does not allow you to break up the score group, 29E5F would allow that had it not been the final round.
Exactly the point I’ve been making. The TD made the right call by disregarding the pairing program, disregarding the 3rd black & pairing the two 4.0 for the final round. That is precisely what the variation intended. If it was round 5 of a 6 round tourney would you have paired the two 4.0 or given them their due color? I would have given them their due color. It specifically states in 27A2, Equal Scores, those exceptions. The reason 29E5f1 was mentioned was to give credence to the TD when he explained why one of them were getting a 3rd black.
If WinTD has this option, I’ve never seen or heard of it, let alone use it. What’s the difference between flipping a coin & having the computer flip a coin? Either way it’s going to randomly decide the fate of one of the players. If you don’t like tossing a coin, give me a better idea for this hypothetical situation.
Are we speaking the same language? 29E5f Says “No player shall be given the same color three times in a row, unless there is no other reasonable way to pair the scoregroup, or unless necessary to equalize colors.” This means that you must pair within the scoregroup, regardless of whether someone gets three in a row.
29E5f1 says “Except in the last round, when it may be necessary to pair the tournament or class leaders, players shall not be assigned the same color three rounds in a row.” This means that a) in rounds other than the last, you must break a scoregroup to avoid three in a row, and b) in the last round, you should not break a scoregroup for this reason.
Aside from saving time, the advantage of having the computer do the toss is that no one can complain about favoritism. I’m not quite clear on what you objection is. Are you opposed to last-round tosses per se? Many directors feel that way, and they simply leave “higher ranked gets his due” in effect. (Of course, if both players are unrated, “higher-ranked” is determined randomly, that is to say by toss.) Personally I think tossing in the last round is slightly fairer, but that’s a complicated argument that really needs a thread of its own.
One of us isn’t. It don’t think it is interpretation. Yes, 29E5f Says “No player shall be given the same color three times in a row, unless there is no other reasonable way to pair the scoregroup, or unless necessary to equalize colors.”
NO PLAYER SHALL BE GIVEN THE SAME COLOR 3 TIMES IN A ROW. I don’t see where the rest of the paragraph applies here.
Where does it say that in 29E5f? Someone is going to have BWBBB. If this was not the last round this would not be a reasonable pairing. I would pair out of the score group in a heartbeat.
Exactly! That is the point. In the last round the TD overturned the computer pairings which DID NOT pair the two 4.0 and he manually paired them, giving one a 3rd black. This was the correct and only decision in the final round.
I agree 100%. I have no objection whatsoever. As I said WinTD does not do this. Since for so long it has been “higher ranked gets his due color” I would have no problem with that. That’s typically the way I handle it. The reason the coin toss came up was because we said hypothetically they had the same rating.
The way I see this is that 29E5f1 is a variation of 29E5f, that is in rounds other than the last the director has the choice of whether to break up the scoregroup or give the third black. In the last round, there is no choice. This is similar to a blitz tournament whether the TD can use rule 3 or 3A regarding illegal moves. In that case, I think that if you’re going to deviate from the standard rule (3) then you should advertise in the pre-tournament publicity, and certainly before the first round. In the case of 29E5f(1), I don’t think that it is necessary to announce that you are using the variation rather than the preferred pairing rule, but it is very clear that either one is acceptable if announced.
Fortunately, this situation doesn’t happen very often.
I’m afraid your interpretation does not match either the wording, or the (extensive) debate on this subject before the rule was changed.
In the 3rd edition of the rules, Tim Redman added a paragraph which stated that “a player may not be given the same color three times in a row.” A number of TDs (including myself) thought that this was just careless wording, and did not mean what it seemed to say, but it turned out that this was really his intention. A majority of the Rules Committee (and the Delegates) did not agree with this rule, because pairing players with equal scores is the first priority of the swiss system (after not pairing people twice, of course). In the 4th edition (actually a bit earlier than that, but it was published in the 4th) the rule was explicitly changed to state that pairing within the scoregroup took precedence over avoiding three in a row. The 29E5F1 variation was added to accomodate peoole who agreed with Tim about the importance of avoiding three in a row, without compromising the basic swiss system too much (thus the last-round exception).
You are dismissing the phrase “unless there is no other reasonable way to pair the scoregroup,” which is actually the point of the paragraph. If you drop the only two players with 4, you are not pairing them within their scoregroup, when it was possible to do so. Unless you are using 29E5f1 or some similar variation, you are simply not allowed to do this. (You can also assign three in a row for equalization of colors, but that 's a different matter, and one that hardly ever comes up.)
In WinTD, under , there is a checkbox for “Due Colors to Higher Ranked.” If you uncheck this, the program will toss for colors in the last round. It will also do some other things you may or may not want. From the Help menu:
Due Color to Higher Ranked
If this box is checked, WinTD will always give the due color to the higher ranked player whenever the two players have an equal claim to a color. (Note that this does not mean the higher-ranked player necessarily gets White–she gets the color that both players would expect to receive). If not, WinTD uses the more “modern” treatment of favoring the higher ranked player in even or plus scorepacks and the lower ranked player in minus scorepacks, and “flipping a coin” in the final round.
This came up once in a tournament where Browne and Gulko had the same rating. Browne insisted that I randomize the seeding each round “… because B will always come before G in the alphabet.” I agreed he had a point, though I doubt it had much effect on the final result.
OK. So getting to the point of this whole debate. Would you or would you not pair the two 4.0 if it was not the final round, knowing they would have BWBBB?
