# Thread: Top v Bottom or Colour

1. ## Top v Bottom or Colour

Spun off from Q 3 in the Doeberl Cup Arbiting Quiz

As I said in the quiz there are no right and wrong answers and I think that this is most certainly the case here. While Charles Z and Cathy Rogers think that Top V Bottom is part of the system, I have always thought that colour takes preference. Having looked at the rules, both on the FIDE website and in Reubens book, it is my opinion that the rules are so badly worded that we could all be right.
Some observations:
At the start of the pairing process you group players into score groups and then sort them into rating order. You then split them into 2 sub-groups, top half and bottom half. <This indicates that top v bottom is part of the system>
In the rules section A9 refers to transpositions (ie re-ordering within a subgroup) and exchanges (swapping players between sub-groups). However this section does not say when to apply them, just a reference to "sound pairings".
In section B2 (Relative Criteria) part (b) it says that "as many players as possible should recieve their colour preference". It does not mention top v bottom as one of the pairing criteria <This indicates that either colour is more important or that top v bottom is so fundamental it wasn't explicitly stated>
In section C6 the method described is to pair the first player in S1 with the first player in S2, the scond in S1 with the second in S2 etc etc. It then states "if p pairings are obtained in complience with B1 and B2 the pairing of this scoregroup is now considered complete". When the rules were first published the clause was badly written leaving it unclear whether you only had to fulfill B1 and not B2. <To me this makes clause B2(b) an integral part of the pairing process>
If p pairings are not obtained you firstly transpose players within S2, repeating step C6 as you go, and if you exhaust all the pairings without succes, you exchange players between S1 and S2 and repeat C6 and C7. <And I think this is the root cause of the confusion. While a computer can mindlessly carry out these steps, for an arbiter to carry out this process exactly as written would take quite some time. Instead humans tend to take shortcuts which may result in a different pairing than the pairing system would arrive at>

Now my reading of the rules would indicate colour is more important than top v bottom but I think there is enough wiggle room in the wording for the opposite conclusion to be also drawn.

But while thinking about this another question occured to me: Under what circumstances would you need to pair within a sub-group? If all of S1 had the same colour preference and all of S2 had the opposite preference then their wouldn't be a problem. Even if both sub-groups wanted the same colour you could still pair top v bottom, with the bottom group simply getting screwed on colour. My guess is that this problem only occurs if there are some topvbottom pairings that cannot occur because the players have already played or the colour balance would end up +3 or -3.

One final observation. On the FIDE web site is the pairing system that preceded the Dutch system. It is listed as a FIDE approved pairing system, so I assume that could be used instead (and indeed it is the system I tend to use for manual pairings). But be warned. In the description is an example of swapping players around to find the best pairings and the example pairs players from the same sub-group together.

2. Shaun, if you get a chance can you post the original SP pairings for the last round and explain why they were overruled.

Edit: Actually it'll probably be too much trouble. I don't suppose you or Cathy or Ian can remember the original pairings down to board 4.

3. One of the problems with the rules as written is the lack of instruction on how to deal with conflicts between section B and other sections. This also comes up when dealing with double-downfloat situations (a known Swiss Perfect issue).

I take it that section B tells you the overarching principles and the other sections including A and C fill in the fine details of how to apply them. This means that section B overrides sections A and C in any case of apparent conflict between them. Actually there is really no conflict if you take this interpretation, only if you take the interpretation that B doesn't override the others.

B clearly says that the colour differences shall be as small as possible so I have always taken that to mean that you exchange between top and bottom halves as a last resort to improve colour allocation. Whether this is how the rules should be written is another question:

Towards the end of a tournament a given score group usually includes both higher-rated players playing badly and lower-rated players playing above themselves. If everyone in a group is playing about the same strength then it makes more sense to deliver colour parity than to worry about seedings which may be somewhat rubbery as an indicator of form on the weekend in question.

I have some degree of bias on this one because I do believe that I have good and bad weekends and that my results vary more from weekend to weekend than sheer chance would indicate (meaning that my score after 6 rounds may be a better predictor of my next result than my rating) - although I haven't tested this view statistically. However I also have no reason to personally care about colour allocation because I score the same with either colour.

4. Originally Posted by kegless
Shaun, if you get a chance can you post the original SP pairings for the last round and explain why they were overruled.
Was it a manual pairing, or invocation of "last round special rules"?

5. Originally Posted by pax
Was it a manual pairing, or invocation of "last round special rules"?
I believe it was a manual override.

6. To answer both Pax and Greg ... (NB I don't have the SP files with me as I type this).
It was changed because of the "last round special rules", which the version of Swiss Perfect I was using did not follow. It had avoided the obvious upfloat (Lane up to Rogers) because Lane had floated up 2 rounds before (IIRC). Now at the start of the tournament we had anounced we were dropping that restriction for all rounds. However even if we hadn't made that anouncement, the pairing was still wrong because under the last round rules that restriction is dropped for players on 50% or over. So in this case it was a bug in the program. And again if I recall correctly it had Bjelobrk floating up, Smerdon v Lane and Canfell v Rej.
(Now corrected to Canfell v Lane, Smerdon v Rej)

7. Originally Posted by shaun
And again if I recall correctly it had Bjelobrk floating up, Smerdon v Lane and Canfell v Rej.
Blimey, keg, are you sure you wanted to know that? That's a slap in the face from the devil of the draw.

8. Originally Posted by pax
Blimey, keg, are you sure you wanted to know that? That's a slap in the face from the devil of the draw.
Well yes but Tomek had just beaten Dazza. I'd be pretty confident with White though.

If SP chooses to upfloat Bjelobrk, why wouldn't it make boards 3 and 4 Canfell-Lane, Smerdon-Rej? Doesn't that solve top half v bottom half as well as colours?

Originally Posted by kegless
...
Random events have been conspiring against me lately to give me a preponderance of Blacks, especially in important games. There was the zonal, the Oz masters, the NSWCA October weekender (where SP wasn't overruled when it would have favoured me for once - there's a whole thread on this here) and now I have 4 out of 6 Blacks in the C of S after preparing with White in round 6, thinking it would be virtually impossible to be Black against Ayvazyan since I was half a point ahead (I subsequently found out that if two players have 3 Bs and 2 Ws, sequence is given a higher priority than rank)....
If you had hung around long enough at end of Round 5 C of S would you have known your draw and colour for Round 6 as is computer generated or is it 2 rounds in one day?

In one of the small comps I put on just to hurry and simplify things I declared that all colours will be decided by "hands behind back" prior to play, I got away with it for a few rounds but then Fred F jacked up. Is it too crude a manner? I thought it was terrific but then I would.

10. Originally Posted by kegless
Well yes but Tomek had just beaten Dazza. I'd be pretty confident with White though.

If SP chooses to upfloat Bjelobrk, why wouldn't it make boards 3 and 4 Canfell-Lane, Smerdon-Rej? Doesn't that solve top half v bottom half as well as colours?
Having just arrived home and repaired Rd7 (without the intervention) the pairings you suggest are in fact the ones it gave.

11. Originally Posted by shaun
To answer both Pax and Greg ... (NB I don't have the SP files with me as I type this).
It was changed because of the "last round special rules", which the version of Swiss Perfect I was using did not follow.
Do you mean that when you asked SP to pair round 7 that even though it asked you the question Do you want to apply special last round pairing rules? (Press F1 for Help) it did not actually apply them, or did it not ask you that question.

BTW for the benefit of others not familiar with SP, SP will only ask that question if the Tournament setup screen has the number of rounds set. If you leave it at the default of 11 rounds it wont ask until you pair round 11.

12. Originally Posted by antichrist
If you had hung around long enough at end of Round 5 C of S would you have known your draw and colour for Round 6 as is computer generated or is it 2 rounds in one day?
Rounds 5 & 6 were on the same day. During the week I'd worked out that if I won Round 5 I would almost certainly play Ayvazyan. As I would be half a point ahead of him and neither of us were going to cop 3 Blacks in a row I thought that rank would be the next priority and I would get my due colour, so I duly prepared with White. My fault for not knowing SP's rules as sequence comes next in this scenario. From memory I had BWBWB while Ayvazyan had WBBWB so he got the White.

In one of the small comps I put on just to hurry and simplify things I declared that all colours will be decided by "hands behind back" prior to play, I got away with it for a few rounds but then Fred F jacked up. Is it too crude a manner? I thought it was terrific but then I would.
Why stop there? Make the players draw out of a hat to see which piece to move.

13. i never worry what color i got . it makes no diff . anyways wheres the proof that at a tourni like ours white wins more than black . in the data bases it does but does it happen too in doeberl .

14. I know it is good to know that our arbiters are looking after our best interests. Now can Shaun or someone actually crystallise the whole problem with SP and tell us in two paragraphs, 1. what should SP do and 2. what doesn't SP do that makes over riding it a necessity.

I suppose that a third question might arise, How is the pairing process supposed to be transparent when arbiters are fiddling with them without explanations? At least if SP is used and never over ridden an unhappily paired player can be told to argue with the computer.

15. Originally Posted by BBlooksee
i never worry what color i got . it makes no diff . anyways wheres the proof that at a tourni like ours white wins more than black . in the data bases it does but does it happen too in doeberl .
Sure, I imagine for a lot of people it doesn't make much difference. But for someone like myself who likes the initiative, I'd much prefer to wind up with White in the last round and try and take someone out.

And spare a thought for Solo. He's quite obviously stronger with White, imo, but he copped two Blacks to finish as well.