In the rules (P. 147) it clarifies that “the arithmetic for interchanges and transpositions applies only to the first natural pairing (…) before any transpositions or interchanges are made, compared to the final pairing after all transpositions and interchanges have been made.”
I added the red comma to make it easier to see the two halves of the calculation. It means that after you have tentatively made all of your changes you check against your starting point to make sure you haven’t violated the 200 or 80 point limits.
You might be able to see it was not blatantly obvious in the rules, but you can’t say it is not in the rules and have that statement be valid. The TD tip even clarifies things for people who make your erroneous assumption and lets them know that they have to calculate from the starting point, not from just the last of a series of transpositions.
I don’t have a problem with somebody pointing out a perceived error. It is conceivable that an intelligent person can misinterpret something. One thing that irks me is when somebody makes a misinterpretation and then uses that misinterpretation to say that the rules are bad or wrong or stupid. I really wish a person making such a statement would first actually understand the rules they are making the statement about.
I hadn’t overlooked that TD tip, but it didn’t make sense to me. How do you evaluate multiple swaps as if they were a single swap from the starting state (i.e. from the natural pairings). Multiple swaps may or may not be equivalent to a single swap. If they are, then fine, you can treat it as a single swap and evaluate the multiple swaps as if they were a single swap. But what if the multiple swaps are not equivalent to a single swap, which will be the case most of the time with multiple swaps?
It doesn’t matter how many intervening steps you took, players started with one opponent and ended up with a different opponent.
If you started with
A-B
C-D
E-F
G-H
and somehow ended up with
A-D
C-H
B-F
E-G
then compare A-B with A-D and B-F and see if either the A-F change or the B-D change is within limits.
Do the same for C-D versus A-D and C-H,
E-F versus E-G and B-F,
G-H versus C-H and E-G.
Remember that the starting point isn’t the raw pairing after initially placing the pairing cards (making an assumption that you’d only worry about checking a manual pairing) but rather after you’ve taken those cards and made any necessary changes to avoid having two people play a second time (and possibly to avoid teammate pairings or family pairings depending on the rules of the event). This step was in the “(…)” section of the rule cited.
This is not obvious at all, and if this is how multiple transpositions are supposed to be evaluated, the rulebook should have said so, and not just left a cryptic comment in a TD Tip that is open to interpretation.
How can one practically implement multiple transpositions? According to you, the transpositions are evaluated not one by one as you do them, but only when you have finished with all the transpositions. If there is an X-Y pairing in the final result, it is acceptable if either Y0 versus Y is acceptable (where Y0 is X’s original opponent) or X0 versus X is acceptable (where X0 is Y’s original opponent). If we take your word for it, if both of those two tests are unacceptable, then something has to be undone. What do you undo? Is it back to the start point for the color transpositions? How do you go about retrying? What is the algorithm? If this is the correct interpretation of the TD Tip, it is not an approach to multiple transpositions which seems to be practically implementable.
As I said, I had read this TD Tip but it was a mystery to me what it is supposed to mean. In fact, if you are right about how to interpret it, it is really just one more reason to get rid of the 80-pt and 200-pt rules.
If you are asking how to implement multiple transpositions in a practical manner, I have to say that it is extremely rare to have multiple transpositions so it isn’t really a practical situation. I did (manually) a seven-round, 50-player inaugural K-3 state championship tournament section with 25 players on one team and was able to have all of the pairings make sense while no member of the 25-player team had to play a teammate more than twice in the seven rounds (zero times would have been perfect but that couldn’t be done without distorting the pairings unreasonably - the 25-player team had only a handful of the top 20 players). I didn’t have to make any extremely complex multiple transpositions and all that I made were easily analyzed.
Off-hand, I’d say that if you have overly complex copiously transposed pairings then you’ve probably done something wrong (even a 20 player, 12+ round tournament that gives pairing headaches shouldn’t have overly complex multiple transpositions).
Several details become relevant when validating round three pairings and referencing the MSA cross-table; they are included below.
Published ratings were used, except for two players choosing to be paired according to an established club practice: “Current published ratings used. Upon player’s request their most recent unofficial rating may be used if it is higher than the published rating.”
Player 2114 showed up shortly after round 1 had started. I paired him against a house player (1885) with some time removed from both clocks to keep on schedule.
Player initials added for reference to the crosstable.
SP 2114 2.0 BW
JS 1866 1.5 xB
TM 1942 1.5 xW
EH 1940 1.0 WB
KM 1885 1.0 WB
NR 1850 1.0 WB
AZ 1957 0.0 BW
HA 1807 0.0 BW
I pair by hand as I do not have a laptop to bring to the club. I have a copy of SwissSys that I use at home to prepare the files to be uploaded to the USCF. Natural pairings result in bad color alternation for all boards. The top two boards are within the same score group (sort of, after players have been dropped) so per 29E5a colors can be corrected with a transposition if the rating difference is 80 points or less. I paired as follows:
1866 (1.5, xB) vs 2114 (2.0, BW)
1940 (1.0, WB) vs 1942 (1.5, xW)
1885 (1.0, WB) vs 1850 (1.0, WB)
1807 (0.0, BW) vs 1957 (0.0, BW)
On a regular basis a club member will bring a laptop and run WinTD to verify my manual pairings (he doesn’t play in the tournaments). I still pair by hand when he is present so that I can better learn the pairing rules, but I check my pairings against the computer. My pairings often match the computer, but when there is a discrepancy I usually change my pairings to match the computer. My mistake is often clear at the time, and if not then I follow up at home to understand why. WinTD gave:
2114 (2.0, BW) vs 1942 (1.5, xW)
1866 (1.5, xB) vs 1940 (1.0, WB)
1885 (1.0, WB) vs 1850 (1.0, WB)
1807 (0.0, BW) vs 1957 (0.0, BW)
Since 1942 and 1866 are with 80 points of each other I guessed that I had placed too little value on the highest-rated 1.5 player being paired against the player who had been dropped from a higher score group (2114). I went with the computer pairings. Later when entering the results to prepare files to upload, SwissSys agreed with my manual pairings. I adjusted 1866 to 1861 (to make a >80 pt difference) and repaired but the result was the same. With some trial and error I found that SwissSys will correct these alternation errors up to 200 points. My understanding of the rule book allows transposition to correct alternation up to 80 points. I think that TD discretion can be used to justify a transposition outside of 80 points in the given case.
I still hold the opinion that it would be better to remove rating limits and give preference to the transposition with the smallest rating difference that fixes the color problem in the score group. In the current example this would mean fixing the colors even if the lower-rated player was 1741 (201 point difference). I don’t see justification for bad colors just because one of the players who earned his way into the score group started with a lower rating.
That still leaves questions around the pairings from WinTD, but I don’t have a copy to double-check. In general when there have been differences between SwissSys and WinTD, WinTD appears to have produced better pairings for my small tournaments. This would be an exception.
Looks like a bit of a headache. Heading into round three you had two 0-2 players due black, so one of them was going to get two whites in a row (and it turned out that one didn’t play round four). You had three 1-1 players all due white, and of the two that were paired against each other the one that received white didn’t play round four while the other now had two blacks in a row.
Heading into round four you had three 1-2 players with one of them having black in rounds two and three, having already played the 1-2 player that had white in round two, and having already played the 1.5-1.5 player (the sole 3-0 was paired with the sole 2.5-0.5).
It would be interesting to know the settings on the WinTD transposition and interchange limits, because your original pairings would have limited the field to a single double-white and a single double-black going into round four.
You ended up in a situation where the only way to avoid a triple-black was to do something like the below silly pairings (avoiding the tournament leaders ever meeting each other and having the sole 3-0 skip over two score groups to find an opponent - granted one of those two groups was populated solely by a player already faced).
NR(1-2) vs SP(3-0)
TM(1.5-1.5) vs JS(2.5-0.5)
EH(1-2) vs AZ(1-2)
1 SP 2114 bww 2.0 W5 W4 ?2
2 TM 1942 -wb 1.5 H- W6 ?1
3 JS 1866 -bw 1.5 H- W7 ?4
4 EH 1940 wbb 1.0 W8 L1 ?3
5 KM 1885 wbw 1.0 L1 W8 ?6
6 NR 1850 wbb 1.0 W7 L2 ?5
7 AZ 1957 bwb 0.0 L6 L3 ?8
8 HA 1807 bww 0.0 L4 L5 ?7
Why WinTD didn’t want to make the color transposition, I don’t know. It would be 76 points, within the 80-point limit. Maybe WinTD viewed that transposition as an interchange – within the 1.5 group, it would switch a player (the only player) in the top half with a player (the only player) in the bottom half.
In small tournaments, I have often advocated transposing colors only to equalize, not merely to alternate. In this case, however, two bad colors (out of four) is easily sufficient to avoid the “camp” effect often produced by too-good colors in round 3.
So, maybe you should have gone with your gut. But you caved in to WinTD, and ended up with round 4 as follows:
1 SP 2114 bwwb 3.0 W5 W4 W2 ?3
2 TM 1942 -wbb 1.5 H- W6 L1 ?7
3 JS 1866 -bww 2.5 H- W7 W4 ?1
4 EH 1940 wbbw 1.0 W8 L1 L3 ?6
5 KM 1885 wbw- 2.0 L1 W8 W6 --
6 NR 1850 wbbb 1.0 W7 L2 L5 ?4
7 AZ 1957 bwbw 1.0 L6 L3 W8 ?2
8 HA 1807 bww- 0.0 L4 L5 L7 --
Given the two drop-outs, at this point the above pairings are pretty much forced – three straight blacks and all.
But, what would have happened with your preferred round 3 pairings? Let’s assume that the higher-rated would win on board one, and that the bottom two boards would come out the same as in the actual pairings. This leads to three possible results on board two, and three possible sets of round 4 pairings:
I would still prefer the original pairings had the rating difference been over 200 points, which is why I support correcting colors within a score group using the smallest rating difference for transpositions/interchanges, but not imposing a limit on point difference.
I would like to highlight that my opinion of no rating point limit to fix colors applies to changes within a score group. In the example above three consecutive blacks is preferable to pairing outside of the score group.
That would be the reason. If there had been two extra (lower rated) players in the 1.5 group, it would make the 76 point switch, but won’t switch just to correct alternation if there are only two.
But you caved in to WinTD, and ended up with round 4 as follows: