Application of Rule 29E5f. Colors in a Series and 27A1

Running Chess Tournaments
1

Greetings - I have a small pairing exercise I thought was interesting. Thanks for any feedback in advance:

1.5 score group after round 4 of 6:
(pairing #, rating, color sequence)
21 1485 WBWB
26 1089 -BWW
28 1049 -W-B
34 747 BW-W
37 591 -WBW
42 261 -BWW

previous pairings:
21 v 34
28 v 34 (forfeit win for 28)
21 v 37

Initially, I noted after pairing 21 with 42, that I would have 3 players strongly due black for alternation and equalization. I would have to pair it as follows:
21 v 42 = 1485 v 261
34 v 26 = 747 v 1089
28 v 37 = 1049 v 591
This would lead to one player (34) getting BW-WW.

The computer disagreed - and I believe it is applying rule 29E5f, which states: “No player shall be assigned the same color three times in a row unless there is no other reasonable way to pair the score group or unless necessary to equalize colors.” Even with his forfeit to break up the 3 Whites, it forced both 34 and 26 into getting black and paired it:
21 v 42 = 1485 v 261
28 v 26 = 1049 v 1089
37 v 34 = 591 v 747

So, I get that - but since we are arriving at the end of the tournament near prize time (u1400 and u1000 at play here), I could see someone raising questions about that score group. For example, both the people vying for the u1400 have to play each other - when the natural pairings would give them each a more favorable matchup. The 261 player, in place for a u1000 prize and given his 1224 point deficit to his opponent, may question why the 591 and 747 were allowed to play instead of also getting paired with higher players.

i had the thought that if I ignored the previous forfeit matchup that I could simply do a normal transposition (26 and 28 are 41 points apart) and pair it like this:
21 v 42 = 1485 v 261
37 v 26 = 591 v 1089
28 v 34 = 1049 v 747

I thought this was some side case - not covered in the rule set - but there it is clearly under the most basic of swiss pairing rules, 27A1: “If two players were paired against each other earlier in the tournament, but the game was forfeited due to the nonappearance of one, they may be paired against each other again”. So… I can pair it in this third way even though SwissSys 9.x does not.

Thoughts? Would you adjust the computer pairings or let them stand? if you were slightly afraid the same player would flake again and no-show, would that impact your decision?

2

Ugh.

No wonder no one has responded so far. The situation seems near impossible.

Since three of these six players have just had 2 whites in a row, it seems absolutely necessary to assign black to all three, and thus pair each against one of the remaining three who have not just had two whites in a row.

As far as I can tell, there are only four ways to accomplish the above, if rematches are to be avoided:

A.
21 WBWB vs 42 xBWW
37 xWBW vs 26 xBWW
28 xWxB vs 34 BWxW

B.
21 WBWB vs 26 xBWW
28 xWxB vs 34 BWxW
37 xWBW vs 42 xBWW

C.
21 WBWB vs 42 xBWW
28 xWxB vs 26 xBWW
37 xWBW vs 34 BWxW

D.
21 WBWB vs 26 xBWW
28 xWxB vs 42 xBWW
37 xWBW vs 34 BWxW

Comments:

A. Ugh. No interchanges, but includes the pseudo-rematch 28 vs 34 which was a forfeit last time.
B. Ugh. Involves an interchange of over 500 points, and includes the pseudo-rematch.
C. Ugh. Involves an interchange of over 300 points, but avoids the pseudo-rematch.
D. Ugh. Involves an interchange of almost 350 points, but avoids the pseudo-rematch.

Ugh. I guess I’d go along with A, as the least of the four extreme evils.

In general, for that very reason and others, I would hate to pair a pseudo-rematch where one player forfeited last time. In this case, though, there does not seem to be a reasonable alternative.

At least the player did show up for round 4 after forfeiting round 3. I assume he gave a reasonable semi-excuse for forfeiting, otherwise he would have not been paired for round 4.

Bill Smythe

3

Thanks much Bill! I was surprised at no one taking it on :slight_smile:

I didn’t see C as an interchange, but glad you did - I saw it as the by-product of the application of 27A1 (no rematches) 29E5f (no consecutive colors 3 times in row unless impossible otherwise). The clause that rematches are ok if the matchup was not actually played was a surprise to me.

This situation actually arose out of the same tournament I posted earlier about where the computer did not drop a player to improve colors in a lower score group. That was concerning round 2. What happened for round 4 was kind of crazy (i thought). I had 9 players in the 1.0 score group. 8 of them were strongly due Black. 1 was weakly due Black. I didn’t post about it - because despite the seemingly rare situation, my pairing and the computer pairing agreed when addressing it.

As for the forfeit - his father did tell me a few days in advance that he would not make it - but a week after the pairings had already been made. I could not get the player to arrange a make-up match with his opponent in the interim - but he did assure me that he would be able to make subsequent rounds, so I put him back in. I guess I am too worried that this situation will repeat itself for this upcoming round.

4

This is a reason not to post pairings far in advance.

Alex Relyea

5

For a 1 game every two week tournament - I like to get the pairings out ASAP so the makeup games can occur before the scheduled date - not after. If they do not occur by the time the round is played, it does two things: 1 - reduces the amount of time for a subsequent makeup/rescheduling in the next round and 2 - makes me worry that it won’t happen. The players are informed at the start of the tournament that if they cannot make the scheduled date after pairings have been made that they are expected to put some effort into making them happen outside of the appointed time.

6

What arrangements have you made for a TD to supervise these make-up games?

7

Whenever two players in the top half are paired against each other, and two others in the bottom half ditto, that means there was an interchange, presumably of the lower of the two top-half players with the higher of the two bottom-half players. So, 26 vs 28, and 34 vs 37, meant that 28 (top half) had been interchanged with 34 (bottom half).

Well, that would explain why you still had so many 1.5’s due black in round 5. I’m wondering if maybe you were a little too determined to make colors work in round 3. I’ve never been a huge fan of trying too hard to make colors work in odd-numbered rounds. It often seems better to accept some bad colors in rounds 3 and 5 (alternation) so as to leave yourself with more available good-color pairings in rounds 4 and 6 (equalization). This is especially true in small events (or small sections). Now, 42 players (I know you had at least 42, because one of the players was numbered 42) doesn’t exactly sound like a small section, but of course within the somewhat larger event there were some small score groups. I’m wondering what would have happened if you had made fewer (perhaps even zero) alternation transpositions in round 3.

Bill Smythe

8

That makes no sense at all, at least ex ante. Whether fixing fewer colors in round 3 than you could would help pair round 4 depends entirely on the results—you’re more likely to produce the three-in-a-row boxouts that generated the OP.

Now, given a set of pairings, it is certainly true that “due color to higher-ranked” is actually not the best procedure when applied to minus score groups, since the player with the color problem is likely to end up in the more extreme (and thus likely smaller) score group. A better procedure would be due color to higher-ranked in the plus scores, to lower-ranked in the minus scores and random assignment to even scores (or even better one randomization followed by alternating high vs low as is done in round one). Of course, given that the current rule is already misunderstood by many—I’ve had GM’s who thought it was higher-rated player gets White—I don’t see that happening.

9

Thanks again guys!

I did have one high-rated player approach me in my first time directing a USCF tournament - seemingly expecting to get white as the higher rated player. As soon as I explained that I believe as the higher rated, he would get his “due” color - which was black - he said something to the effect of "ok… as long as you are consistent and it happens that way every round ".

I had noticed that the favoring the higher rated player for due color does seem to stick me with the same choice for a few rounds.

It appears SwissSys does have a mechanism, looking at its name , I thought it would do as you suggest. Instead it appears to alternate who gets their color due in alternating fashion. I do not know if this means alternating each score group (all highers get their due color in top score group, all lowers get their due rating in 2nd score group, etc), or alternating down the line per pairing.

SwissSys Help: Plus/minus color variation -- This is taken from the USCF handbook, It involves alternating which player gets the color due: First the higher ranked, then the lower, then the higher, etc. In earlier version of SwissSys it was referred to simply as "color assignment variation."

Bill:
As to trying hard with round 3 pairings - I didnt try much at all. I decided to drop a different player into the 1.0 score group (previous thread in these forums), but followed the computers pairings for everything else. As Tom guessed - it was the results that seemed to screw things up - I had three >500 point upsets by players that had 1.0 scores with black in that round. so that left 3 additional players with white in the 1.0 score group that had no business being there :slight_smile:

-Jere

10

Remember, it’s higher-ranked, not higher-rated. Rank is score first, rating second. What does “stick me with the same choice” mean?

That strikes me as a throwaway idea in an old rule book which deserves to be thrown away permanently. That would mean that if you have to pair WB vs WB on board one and BW vs BW on board two, then the top half player would get W in both cases. If the color problems were all in the same direction, that would be one thing (and in a large score group that would usually be the situation), but that’s something that you have to decide in advance to use (players will go ballistic if you apply it on a score group by score group or round by round basis) and so it could make matters much worse in some situations while making it only marginally better in others.

11

Thanks Tom.

Hmm thinking on it more - i guess its just chance - having 2 top players with same color history - preferring the higher ranked (yes, not rated - I did have that come up once with a dropped player with same color history as his opponent!) - and then having to do the same the next round - also having to prefer the higher ranked again. This situation would obviously happen often early in tournaments where there aren’t a lot of color history variations.

Thanks for the reminder about announcing some of these variations in advance. I forget I am not doing this “in a vacuum” sometimes. :slight_smile:

Oh and answering Ken’s question about TD supervision. When possible I have encouraged players to attend the weekly club meeting where I would be present. In other cases - is there no leeway available to trust the players to have a sporting/fair match? One of my first USCF games and a most enjoyable match was in this situation - but perhaps it was enjoyable because it was a memorable one - in the middle of a snowstorm, I was lucky enough to get my first win vs. a class A player!

12

Yes, a match. However things are different for a game as a part of a tournament. This is more or less the same issue as having players play remotely (likely online today, but also by phone or fax) without TD supervision.

Alex Relyea

13

From the Official Rules of Chess:

14

What you say about the best way to allocate due color makes sense. Has the ratings committee given thought to proposing a rules change to that effect?

15

I think you meant “Rules Committee”.

16

I did indeed. Thank you for the correction.

17

So… after all that, I did as suggested and decided to not pair the players against each other that previously had a forfeited match (allowable under rule 27A1). The player who showed last time and got the forfeit win did not show this time. I was still satisfied with the decision (hindsight is 20/20) - as it didnt give a full point to a player who had previously no-showed, but it was still ironic i guess.

I dont think I will have time to check back here before sending out pairings. But Ive been sliding down that “slippery slope” I mentioned previously - starting to doubt all computer pairings.

Given the following 2.5 (out of 5) score group:
4 1895 B WBWBW
20 1485 B WBWBW
24 1167 W WBBWB
25 1089 B -BWW-
26 1052 w -WBWB
34 747 w BW-WB

so… I’ve got 3 due W (1 strongly) and 3 due Black (all strongly)… here are the illegal matchups:
4-24 (round 5 draw)
25-26 (siblings)
20-26 (round 4 26 won)
20-34 (round 2 draw)

how would you pair it?

I used 1 transposition and 1 interchange under the 200 point guideline to come up with:
26 vs 4 (26 transposition with 25)
24 vs 20 (24 interchange with 25)
34 vs 25

SwissSys did not agree and actually appeared to use simpler logic by eliminating the illegal matchup first by transpositioning 26 and 26 (same as I did). But then it let all 3 pairings stand as follows:
26 vs 4
20 vs 25
24 vs 34

which as 1 problem for equalization (20) and 1 problem of alternation (34)

I believe I am probably not understanding something. Assuming I also eliminate the illegal matchup first (so it wouldnt count as a transposition), I would be left with the following 2 pairings:
W: 20 1485 B WBWBW
B: 25 1089 B -BWW-

W: 24 1167 W WBBWB
B: 34 747 w BW-WB

why shouldn’t I interchange 25 with 24 ?

Like I said earlier, I am going to have to send out pairings, but I am sure your answers will enlighten me… or push me further down the slope :slight_smile:

Thanks in advance
-Jere

18

I’ll look at your post more fully in a few minutes, but for now I’ll make a few general comments.

Hmm – in that case it would have been poetic justice to have paired them against each other again. The player who forfeited last time would now find out how it feels to sit around waiting for a no-show opponent.

In fact, you could keep pairing them against each other every round, until they finally both show up on the same night. :exclamation:

There is actually a benefit to sliding down that particular slippery slope. It teaches you how to make pairings. Entirely too many TDs these days simply rip the pairings off the printer and post them without even looking. Who knows, someday one of these TDs may even need to know how to make pairings, and then they’ll be up the paddle without a creek.

Bill Smythe

19

Assuming that you didn’t miss any other pairing blocks, those look to be within the rules. 20 playing 24 is the most “distant” pairing from natural, and that’s 115 points with an interchange. There’s pretty much no other way to fix the colors (20 having no other potential opponent due White) and since there’s an equalization problem otherwise, that’s within the limits.

20

Indeed, why not? By making an interchange of a mere 78 points, together with a transposition well within the 200-point limit (in fact, it looks like 37 points to me), you have eliminated all color equalization issues – and all color alternation issues too, for that matter.

I like your pairings better.

I must admit, though, that I don’t care for the idea of eliminating the horrible pairings first, then making improvements as a second step. For example, what if eliminating the horrible pairing involves a 300-point transposition, and then the second-step improvement involves a 260-point transposition in the opposite direction for the same player? Then the net transposition for that player is only 40 points, yet this pairing option might be overlooked by a TD (or program) that is considering only the absolute differences at each step.

Much better, I think, is to compare the raw pairings (very raw, even pairing the same players twice) against the final proposed pairings, rather than comparing the intermediate pairings against either the raw or the final pairings.

In this case the raw pairings are:

4 vs 25 (bad equalization for player 4)
20 vs 26 (already played each other)
24 vs 34 (bad alternation for player 34)

(Note: In the above, the higher-rated player (rather than the player being assigned white) is listed on the left, and the lower-rated on the right.)

The proposed pairings are:

4 vs 26
20 vs 24
25 vs 34

Now let’s list, for each player, the rating difference between that player’s raw opponent and his proposed opponent. Let’s also list, for each player, the problem (if any) that is solved for that player by making the proposed pairing rather than the raw pairing:

4 vs 26 (for 4, a 37-point switch to equalize colors) (for 26, a 410-point switch to avoid a repeat pairing)
20 vs 24 (for 20, a 115-point switch to avoid a repeat pairing) (for 24, a 738-point switch to avoid no problem)
25 vs 34 (for 25, an 1148-point switch to avoid no problem) (for 34, a 78-point switch to alternate colors)

Now, are these transpositions acceptable? How do they compare with the rules for evaluating a simple transposition?

A simple transposition (involving just two pairings) is acceptable if it is within the limits for solving the problem (e.g. 80 points for alternation, 200 points for equalization, 9999 points to avoid a repeat pairing). In deciding this question, the smaller rating difference is used. For example, to improve color alternation, Jones may be switched with Smith provided that either Jones and Smith are within 80 points of each other or Jones’ and Smith’s opponents are within 80 points of each other.

The analogy to the above for a compound transposition is (or at least should be, IMHO) that a compound transposition is acceptable if, for each individual pairing in the compound transposition, the consequences are acceptable for at least one of the two players in that pairing.

If you examine the proposed pairings, you’ll see that they pass muster. Go for it!

Bill Smythe