Right now, we're discussing this in another forum, and I'd love to have some clarification about this because the MTR don't exactly explain this.

When running a tournament in swiss-mode, are the tie-breakers used for every round or do WER/DCIR pair the players with the same number of match points randomly against each other?

It's random within points - at least, as random as a computer algorithm can accomplish.

Originally posted by Scott Marshall:

It's random within points - at least, as random as a computer algorithm can accomplish.

Thanks! Are the tie-breakers used for the pairings for the last round?


And do we know about any specific reason for this random method? I actually was under the assumption that the tie-breakers are used every round because the WER calculates them anyway.

I've heard that WER uses standings (thus, tiebreakers) to pair the final round.

That implies an important caution for scorekeepers: make sure you have the number of rounds correct before you get to the next-to-last round! :)

Just ran some checks (as I haven't for a while):

DCIR and WER behave identically here: the final round is paired by standings, while the pairing for other rounds is random Swiss.

I believe that, at one point, WER operated differently: the first round was pure random, all other rounds by standings. I remember noticing this in some testing I had done, and also remember it as being the reason there were a lot of “chain pair-downs” at some events, where a pair-down was continuously getting paired down for all future rounds. (which, now that I think about it, I haven't seen in a while)

Can someone else confirm that the chain pair-downs are gone? I've noticed
them happen in the past, but I haven't been particularly watchful recently
and I don't run events where pair downs are common enough for me to notice.


2013/11/21 Jordan Baker <forum-7161-e959@apps.magicjudges.org>

WER now has an option to control whether it pairs by standing or randomly, I believe. The default is as Jordan described. Chain pair-downs will still occur if you pair every round by standing, as a paired-down player will usually have worse breakers and so end up in the pairdown slot in the next round.

Originally posted by Sean Hunt:

WER now has an option to control whether it pairs by standing or randomly, I believe. The default is as Jordan described. Chain pair-downs will still occur if you pair every round by standing, as a paired-down player will usually have worse breakers and so end up in the pairdown slot in the next round.

That option is for how it seats players, either by standing (top players are at top tables) or random. (players are spread out everywhere) The decision of using one instead of the other is more deserving of a different thread :)

This may on tangentially related, but is it possible WER uses players by name to break ties at any point? I've had a few players at FNM complaining that they always seem to play each other, and I noticed that the people complaining always have similar last names.

Our FNM is small enough this could just be chance though, but I would love to give them a definitive yes or no :)

I'm relatively sure player names don't figure. What might is order of entry
into tournament. I suspect it's just chance though. I remember back when my
local FNM used to be small, I played against the same person round 2 three
or four weeks in a row.


2013/12/3 Matthew Turnbull <forum-7161-e959@apps.magicjudges.org>

A local store runs 60-70 player prereleases and does a morning and afternoon one with a round cap. For the last two sets, I have had the same round one opponent in both the morning and the afternoon, despite maybe half the players in the tournament changing! We have some suspicions about the R1 random pairing…

I have had this experience so many times I've given up pointing it out and sending in WER reports. This same problem also creates the situation where certain players will get paired down or paired up whenever it comes up that they might be in contention. Somewhere in the WER algorithm, there is something happening when it pretends to randomize that causes it not to - whether it is using alphabetization, DCI number order, or the order in which they were entered into the local player database, something is going on that makes pairings not at all approaching random (close enough for machines and people to perceive of it as functionally random) but what is happening within the code is arcane and hidden from the public.

For small tournaments do you think it would be good customer service to randomize it ourselves? Obviously this becomes a waste of time quickly, but when possible it could be worthwhile.

I recall seeing official word that where WER (or it might have been DCIR at the time) differs from written policy it can differ in ways that the PTB want it to differ. I think this recollection was in regard to having the final round of swiss not paired randomly so it might not apply to this perfectly but just because WER does something different to how the MTR says it should be done doesn't mean that the behaviour is unwanted.

That is clearly true on some level - and really does beg the question of why the official docs are not just updated. But much of the unwanted behaviour we are noting here is truly unwanted, not just possibly stealth official, but simply bad.