I did a brief (read as: not very thorough) search of the forums and couldn't find this:
I know how SOS and SODOS are calculated to determine who is the eventual winner of a Swiss System tournament.  But what is the process to decide who advances to which game during the tourney?

For instance in this tourney:
http://www.wargear.net/tournaments/view/641

I understand how hoterpone and I both ended up in the "championship" game (of sorts) but what metric was used to pick Razing Aces and Quazimoto?
It can't be SOS or SODOS, because either way, Mad Max should have made it into the finals then.
Is it just random assignment of the who of the 12 players who have records of 1-1?

i believe that was the case; but can't recall for sure.

Ahhh - it is quite possible that we only use those measures to break ties and not to determine seating. How do we determine seating?

Randomly I believe.

I'm not so good at the maths, but I'd think that would increase the likelihood of ties, where if you used the tie breakers to determine seating in the successive rounds, those tie breaker metrics would have a tendency to diverge.

No?

BorisTheFrugal wrote:

It can't be SOS or SODOS, because either way, Mad Max should have made it into the finals then.
Is it just random assignment of the who of the 12 players who have records of 1-1?

Yes, exactly. Is there a better (fairer) way to do it)?

This might be the opposite of the way some would suggest using the SOS or SOSOS when scheduling, but I'm thinking of an odd way to guarantee there are no ties come the final round. Basically, always use the SOS/SOSOS to eliminate the odd man out.

At it's simplest, consider a Swiss where three players have 3-1 records going into the final round.  Just pair the top two by SOS. 

Backing this paradigm in to previous rounds, pairings in all games would be based on SOS, always pitting #1 against #2, 3-4, 5-6, etc..  Over time, the SOS's (and SOSOS's) would diverge pretty dramatically, and the winner will not only have had the most difficult schedule, but also will always play the #2 player in the final.

When there are SOS ties (which I think there would be in earlier rounds), the SOSOS would be consulted using the same paradigm.

Disclaimer:  I come up with these zany ideas pretty early in the morning, then I have a a cup of coffee and reality usually sets in.

So isn't this just the same thing as how the overall winner is determined now?  Makes sense to have it determine the top players from each round.

The only benefit I see from a random selection is the concept of a Wildcard.  Say you have 7 players with the same record after one round in 4-player matches.  The 8th slot gets chosen from the next-best record and becomes the Wildcard. 

AttilaTheHun wrote:

So isn't this just the same thing as how the overall winner is determined now?  Makes sense to have it determine the top players from each round.

I'm not suggesting changing the way the overall winner is determined. I'm suggesting changing the way that players are matched against each other in later rounds.  Currently I believe it is random.

Currently, if 3 players enter the last round with the same record, the final match-up is randomly selected with one player as odd man out.

Consider (and I'm pulling these numbers out of my butt).

        SOS         SOSOS

A        9               27

B        8               25

C        7               23

Let's say A is randomly matched against C and B plays someone with an inferior record.

C beats A, and B wins his match.  It's  possible for B and C to end up with the same SOS and SOSOS's.

Using my criteria A would always play B in the final round.  If B and C both win their matches, B will win the tournament having clearly played the tougher schedule.

BorisTheFrugal wrote:

I'm not so good at the maths, but I'd think that would increase the likelihood of ties, where if you used the tie breakers to determine seating in the successive rounds, those tie breaker metrics would have a tendency to diverge.

...

It sounds like BTF and M57 are saying the same thing here, no?  

Concept being that the SOS and SOSOS tiebreakers be used to not only determine overall tournament winner but also be used to determine seating for successive rounds.

Makes sense to do!

AttilaTheHun wrote:

BorisTheFrugal wrote:

I'm not so good at the maths, but I'd think that would increase the likelihood of ties, where if you used the tie breakers to determine seating in the successive rounds, those tie breaker metrics would have a tendency to diverge.

It sounds like BTF and M57 are saying the same thing here, no?  

I sensed that though we were saying the same thing in terms of using SOS and SOSOS to determine seating (really it's his idea), we might have been thinking the opposite regarding how that might be accomplished. Some might be tempted to pair the lowest against the highest to create the divergence BTF speaks of.  I worry that method would backfire whenever the underdogs win.

That sounds like a good approach to me M57.

I'd be curious what Hugh and Amidon think about the solution.  I haven't tried modelling it to see if I can break it, but on the surface it looks like it could be air-tight.

The proposal is to use SOS, then SOSOS, then Random, for seating? Matching the top brackets in succession? That's reasonable.

Hugh wrote:

The proposal is to use SOS, then SOSOS, then Random, for seating? Matching the top brackets in succession? That's reasonable.

Yes.

 ...but does it guarantee a clear winner, without the need for a tie-breaker beyond SOSOS?

M57 wrote:

Hugh wrote:

The proposal is to use SOS, then SOSOS, then Random, for seating? Matching the top brackets in succession? That's reasonable.

Yes.

 ...but does it guarantee a clear winner, without the need for a tie-breaker beyond SOSOS?

Have we had a Swiss tie yet? The heuristic of hardest schedule seems to minimize the likelihood. With computer experiments, we could either exhaust possibilities in a lot of cases or find a counterexample. (The structure of round robins allows for clean arguments/examples, but not so for Swiss.)

Anyway, open question, left as an exercise to the audience.

M57 wrote:

Hugh wrote:

The proposal is to use SOS, then SOSOS, then Random, for seating? Matching the top brackets in succession? That's reasonable.

Yes.

 ...but does it guarantee a clear winner, without the need for a tie-breaker beyond SOSOS?

Point of clarification - I think Hugh's restatment is missing a step, but he is correct for the sake of this argument:  The proposal is to use Tournament Record as the first determining factor, then use SOS, SOSOS, and then Random.

BTF's edit covers it.

I'm just thinking out loud here. Using this method, if a player who has lost in earlier rounds to someone "in the running" continues to lose, they could potentially push that winning player out of the running.  Put another way, once you win a game against someone, generally speaking, you want them to do well coming down the stretch in order to keep your secondary and tertiary numbers high.  Therefore, you are potentially at a disadvantage when you are pitted against the weakest of players in the early rounds.

I don't have a problem with this; it seems to be worth the up-side - as long as the first round is a completely random draw.

Come to think of it, this is not really different than the current method.

M57:  2 thoughts:

1) I think that it doesn't matter if they lose on their 2nd round, because the seat assignment for 2nd round games (using SOS and SOSOS) will be chosen prior to them losing their 2nd round game, because Swiss System isn't played multiple rounds at a time.  This will come into play for the 3rd round, though, in which case....

2)  If the loser player loses subsequent rounds, then they are actually a bad player (relative to the rest of the tournament players).  As such, the "in the running" player's ranking SHOULD slip down the rankings because that win he got in Round 1 is technically worth less (because it was against a worse player than originally expected).

I don't care for how Swiss tourneys are resolved now so I don't play them.   SOSOSOSOSOS makes sense to me for round robin since everyone plays one another, but I don't like the idea of losing a tie-breaker in Swiss because I randomly drew a "weaker" schedule. I also don't enjoy playing out tournament games for which I have no chance of winning the tourney.

I think Swiss should be melded with an elimination type system.   (Basically if, after playing out the regular number of swiss games, there is a tie then more games are played by the top players until there is no longer a tie.)