This past weekend at Game Day I had an interesting situation come up, during round 2 a player received a by. Then in round 4 while they and two other plplayers had the lowest points at 3 the player was awarded another bye by WER, with our Game Day being a low turn out this meant that the player would make the top 8 without actually winning a played match.

Rightfully so some of the players were a bit upset by this and asked if I could just repair the round so that the player wasn't awarded the bye after having a quick look at the IPG and a search for an explanation of the Swiss Algorithm I couldn't find anywhere that states a player can only receive one bye per event so wasn't comfortable that I have the right to change the pairings and explained that the software chose the pairings and I wasn't allowed to change them.

Is a player allowed to be awarded more than one bye due to uneven numbers? Am I allowed to repair a round to avoid it?

It's very unusual, but it can happen.

Manually manipulating pairings to get a “more desirable” result - no matter what criteria you use to define “desirable”, nor who it benefits or hurts - is likely to be considered Tournament Fraud. Don't do that.

d:^D

This got some discussion on the UK message boards, on the grounds that Swiss pairings don't allow for a second bye, and as such the WER behavior is incorrect (i.e. in contravention of the MTR).
(MODO behavior doesn't trump the CR, so why should WER behavior trump the MTR? ;) )

If this is permissible in Magic, can we get an up-to-date reference/document explaining what the Magic Swiss system is and how it differs from “standard” Swiss algorithms? There doesn't seem to have been such a thing for years (most recent I could find was from 1999).

Edited James Winward-Stuart (Feb. 18, 2015 12:43:14 PM)

I have a feeling that it's a small flaw in the logic used to produce the
modified Swiss pairings in WER. I have attempted in the past to sit down
and create my own algorithm for producing pairings for Wargaming systems
(like Warhammer and Flames of War), and been vexed by the what-ifs. To do
the programming, you have to create an intricate tree of logical decision
making. What do you do when this situation happens? What about with this
small wrinkle? And you have to do that for all kinds of situations and
often it creates situations where you have to start back at the top, but
change a switch so that it goes down a different branch of decisions.
Sometimes what ends up happening is that your decision tree can only end up
with one illogical outcome such as giving a player a 2nd bye, because no
other path through the decision tree works to create pairings.

Sometimes it's easy as a human to look at something and go “That's wrong,”
but when you're trying to do the same thing with logic, it can be a very
perplexing proposition to get it right in every single possible scenario. I
do believe that it is made significantly more difficult by a few factors
such as small number of players, irregular ties, and an abundance of
dropped players.

It may be better to say that tournament operations should be handled by how WER spits out data, or, put another way, the behavior of WER at any given moment is proper procedure, independent of any documentation. If it rounds a tiebreaker incorrectly, or it double-byes a player, or pairs a player against the same person twice in two consecutive rounds, or pairs a player down when a pair-down was not required, you still go by what WER says, as doing otherwise means that two identical tournaments in different regions are running a standard (the term, not the format) WPN tournament differently, because one “caught” something while the other didn't.

(you may be able to guess that I've seen these happen a few times…)

There may also be the chance (likely not in this case, but just as something to keep in mind in general) that you're positive X behavior is incorrect in the moment, and fifteen minutes later you realize that it was intentional for XYZ reason. (it's happened to me more than once when studying DCIR/WER)

Report a bug if the behavior isn't correct, but following WER's behavior is about the only way to guarantee consistent events, even if the consistency isn't what some would consider ideal.

Also, for completeness sake, note that there are random reasons why a player could receive more than one bye in a 100%-standard Swiss Magic tournament, however these tend to require either DQs or formats involving multiple draft pods to happen.

Originally posted by George FitzGerald:

Sometimes what ends up happening is that your decision tree can only end up
with one illogical outcome such as giving a player a 2nd bye, because no
other path through the decision tree works to create pairings.

Good point - perhaps what we're seeing is the best of a set of bad options. But I've used WER too often to trust that that's what's going on without some deeper investigation - and if the deeper investigation does reveal that it was an error on the part of WER, what should we do then?

Jordan Baker
It may be better to say that tournament operations should be handled by how WER spits out data, or, put another way, the behavior of WER at any given moment is proper procedure, independent of any documentation. If it rounds a tiebreaker incorrectly, or it double-byes a player, or pairs a player against the same person twice in two consecutive rounds, or pairs a player down when a pair-down was not required, you still go by what WER says, as doing otherwise means that two identical tournaments in different regions are running a standard (the term, not the format) WPN tournament differently, because one “caught” something while the other didn't.

If the goal is to keep tournaments identical, then keeping to WER's bug-of-the-day rather than the rules seems to me a bad approach. Perhaps we could follow a protocol like “try unpairing and automatically repairing once to see if it was a one-off error, but otherwise stick with what WER gives” but that seems awkward - and it's still hard to explain to players when the situation is an obvious bug (the case here isn't necessarily a bug, but our approach to WER doing odd things has to account for the fact that it is unstable, buggy software, that sometimes does do things that are unambiguously wrong).

Jordan Baker
note that there are random reasons why a player could receive more than one bye in a 100%-standard Swiss Magic tournament

I can't think of any legal reasons why a player should receive more than one bye, given that such an event is (pending an update on the Magic Swiss Algorithm) not permissible in the Swiss pairing system and hence not allowed by the MTR.

Edited James Winward-Stuart (Feb. 18, 2015 01:47:44 PM)

James, I didn't say that a second bye is permissible; neither did I say that it wasn't. What I said, essentially, was “don't muck with the pairings just because you think you know better”. (You don't. See Jordan's post, above.)

If you discover an incorrectly entered result, then manual pairings may be required after correcting the result. If you have fixed seating, you'll have to manually swap two pairings between the fixed & non-fixed tables. If WER or the PC crashed, you may have to re-enter pairings manually. And if something crashed, you may even have to resort to manual pairings, then enter them later.

But don't commit tournament fraud by re-pairing until the results meet your criteria.

d:^D

I have observed this several times. It usually happens in the last round of
the tournament, because in the last round WER uses not only players' match
points, but also their tie-breakers when creating pairings and pairing
players as close as possible (from tha standings point of view) to each
other is more important for it than avoiding double-byes.

2015-02-18 22:53 GMT+01:00 Scott Marshall <

In the situations I've had this occur, when I've looked more closely at the pairings, the reason that WER awarded the second bye was because there were no other legal pairings. Basically, a small turnout, combined with a couple of drops makes it so that in the last round the remaining players can only be paired by giving the player a second bye because all other pairings/byes result in two players matched who have already played each other. It's possible there's another bug that exists, but I've had this happen about 3 times in the last 2 years, and every time, that was the reason.

I've had a 9 player draft in which WER tried to give me two byes out of three rounds (bye-lose-bye). It's horrendous customer service to have someone turn up to play magic and not to let them play two thirds of the event they've paid to enter. At competitive REL there are more tournament integrity concerns, and I'm not suggesting that any other ‘dislike’ of pairings should be manually altered. However, there is no way at a regular event I'm giving a player a second bye under those circumstances. It's terrible for them and pretty bad for all the other 2-1s who lose out on tiebreakers to someone who didn't actually play any magic.

Edited Matthew Johnson (Feb. 19, 2015 06:14:00 AM)

So to be clear, the outcome of this discussion is: “WER is always right, even when it's wrong”? If this is policy, can this be written into the MTR?

Please, read my posts carefully.

I am not stating policy about WER, I'm cautioning everyone against the dangers of manipulating pairings to achieve a “better” result. The risks include having the event invalidated, as well as being investigated for Tournament Fraud.

This isn't a change to MTR or any other policy doc, because it's simple, common sense.

I think we've exhausted the possibilities for this thread to continue being beneficial…