On Tournaments or A Humble Suggestion for GG2017

[tomjoad]

As we near the final quarter of 2016, I'd like to engage everybody in some thought about what I feel is the easiest way Gaining Grounds 2017 can be improved over its predecessors.

I'll begin by saying that I have really enjoyed GG2016. The new strategies from 2015 are at least as good, and generally are better, than the originals, and I think the 2016 schemes are far far superior to the book schemes. The lessening of the reliance on scheme markers to score has opened up some really cool design space for the Masters in RoF; the 'always hidden' aspect of them is more fun than the originals; the fact that it's slightly harder to score a full 10 points adds tension to so many decisions...I'm a big fan. I'm also glad that some foggier areas of the rules were cleared up and the game seems to be moving in all the right directions.

Obviously, I have one major bone of contention, though. I will illustrate it thusly:

In an 8 player tournament, excluding the possibility of a draw, we will see the standings advance from 8 players with the same record, to 4 players who are 1-0, then 2 players who are 2-0, and then 1 person will be 3-0 and be the winner. We will also have, at this point, 3 players with a 2-1 record, and it with them that I am concerned.

                                   

                                                           

So, looking here, we see Abby and Edgar battling in the finals, and we have Abby emerging as champion. Edgar, we assume, should be 2nd place, right? But this isn't the case most often, and we all know why.

                                   

When we look at the differentials, we see a different story. Edgar won a couple hard fought games, then got beat up by Abby a bit, and has a +1vp diff. Buster lost a round, then got to wallop a newer player in round 2, and with a solid win in round 3 his diff is +7vp. Essentially, by losing in the 1st round, Buster has been rewarded with second place. Edgar faced the, very likely, superior competition on the day, and will finish in 3rd or 4th place because of it. Is that how we want tie breaks to be decided? I hope not, and I come suggesting a better way.

Look above and you'll see that, in the first set of brackets, there is another easy stat we can track and which would serve as a good stand-in to track strength of schedule. By counting the total wins of all your opponents, you can approximate how well they did on the day, and you can give credit to the players who had to face other players who were more on top of their game. So, Buster played Abby (3-0), Donald (0-3) and Fiona (1-2) for a score of 4. Edgar played Abby as well, plus Fiona (1-2) and Gil (1-2) for a score of 5. Edgar gets 2nd place. In this system, vp diff could still be used as the second tie-breaker. For example, if Donald had rallied and beaten Hector in the last round, both Buster and Edgar would have had a score of 5 and Buster would have taken 2nd place anyway due to his superior differential.

I think using strength of schedule (SOS) will lead to the more deserving players doing well more often than diff, and I hope you can agree, and I hope Wyrd will consider this when they finalize their plans for Gaining Grounds 2017.

[Paddywhack]

Interesting idea. I admit it might favor me, as in one tournament I won all my games, but only by 1 or 2 VP, so never finished very high overall. It did feel slightly annoying, but I had fun playing anyway  

Something like this could help prevent that veteran player who randomly gets a new player first round and racks up a huge Diff, putting them far ahead of others. Not 100% sure it's needed, but it never hurts to throw out the ideas.

[Yipikayey]

The problem i see with the system is more about how small the system is. Win/Draw/Loss with a 10 VP potential differential can swing too much the tie breakers in favor of random luck. The problem i see in your example is that it is at 3 rounds with very little people. 

You are on the extreme where it is the all winner take it all. 4 Rounds and you would get Edgar facing Buster for example, and potentially also tying up with Abby if she screws up her last round. Also you are taking for granted that Edgar fought better players than  Buster, which is not neccessarily the case. I mean, it is an example made to suit your point. I acknowledge it can happen, but the oppossite can be true aswell (hard fought battles because he was a worse player than Buster).

The system itself, would be quite cumberstone to manage while keeping swiss pairings. Not impossible, just annoying for the little reward which suffers from the same very issues.

[monkeyking68]

There is also the unintended side-effect of people dropping mid-tournament.  Edgar's opponents drop after losing causing his SOS to plummet.  In the meanwhile Buster's opponents are playing for pride and one of them won.  Buster still edges Edgar.

[ScottishMetaLew]

I'll be honest, I feel SOS is possibly the worst way to deal with things, you're punishing people for something totally out of their control on later games. For instance if I win 10-5,  then next round my opponent loses 10-0 my Sos bombs. It's not my fault he loses 10-0 but it sure as hell could cost me a podium. No league or tournament that I know of for professional sport uses a sos system. It's a punishing tiebreak system that causes more.issues then solves I feel.

[Squirrelich]

Would any of your issues be solved if VP diff was the primary score (for match making too) and win/loss or SOS the tie breakers?

[santaclaws01]

 

Would any of your issues be solved if VP diff was the primary score (for match making too) and win/loss or SOS the tie breakers?

Making it based on VP diff first is definitely not the way to go. Someone shouldn't be able to only win 1 game and still win the tournament.

[Twg]

The current system, whilst flawed is fairly simple no matter what the numbers. A 24-30 player tournament would really push your system to the limits. The 128 players at the nationals would make it impossible.

[Squirrelich]

 

Making it based on VP diff first is definitely not the way to go. Someone shouldn't be able to only win 1 game and still win the tournament.

Touche, but the the 20-0 system (same spread as Malifaux's -10 -> 10 VP system, just no negative values) was popular in the Midwest for WHFB for as long as I remember, and I can not think think of a single tourney where that occurred.

 

Obviously they are different game systems, maybe it is more likely to happen in this one. So no worries

[tomjoad]

 

The current system, whilst flawed is fairly simple no matter what the numbers. A 24-30 player tournament would really push your system to the limits. The 128 players at the nationals would make it impossible.

Why? If you're tracking the pairings, this will be minimal added work. I actually think this is far MORE useful in large tournaments because you are far more likely to be breaking a tie for 1st place, and I definitely would prefer that the national champion be determined by SoS, rather than diff. Both can be influenced by luck, but I genuinely think SoS will be a better indicator of skill, and that this indication will be stronger and stronger the more rounds an event runs.

Also, yeah, I am completely off-board with any system that doesn't use win-loss-draw as the primary determinant of who wins. Wins are wins, and that comes first. The only debate I'm interested in is how best to value those wins.

[solkan]

 

Why? If you're tracking the pairings, this will be minimal added work. I actually think this is far MORE useful in large tournaments because you are far more likely to be breaking a tie for 1st place, and I definitely would prefer that the national champion be determined by SoS, rather than diff. Both can be influenced by luck, but I genuinely think SoS will be a better indicator of skill, and that this indication will be stronger and stronger the more rounds an event runs.

Also, yeah, I am completely off-board with any system that doesn't use win-loss-draw as the primary determinant of who wins. Wins are wins, and that comes first. The only debate I'm interested in is how best to value those wins.

Bob wins all three of his games, so does Susan.  Because you paired Bob against the new player, Bob loses.

Which points out the biggest problem with strength of schedule:  You now have a legitimate reason for a good player to complain (and possibly just walk out) if they discover that they've been paired with a weaker player.

That's the amount of extra work that you're missing:  In order to be fair, the initial pairings have to fair, too.  And getting the initial pairings fair requires something other than just randomly assigning people against each other.

 

[tomjoad]

 

Bob wins all three of his games, so does Susan.  Because you paired Bob against the new player, Bob loses.

Which points out the biggest problem with strength of schedule:  You now have a legitimate reason for a good player to complain (and possibly just walk out) if they discover that they've been paired with a weaker player.

That's the amount of extra work that you're missing:  In order to be fair, the initial pairings have to fair, too.  And getting the initial pairings fair requires something other than just randomly assigning people against each other.

 

As opposed to the current system, in which Bob wins because Susan had to play a better player. Bob is being rewarded for not being challenged. You cannot seriously tell me that's a superior option.

Like I said, both versions have a luck aspect, so we have to decide if it's better to reward somebody for smashing noobs, or for slogging through more even match ups. To me, there is no question.

[Kadeton]

 

Bob wins all three of his games, so does Susan.  Because you paired Bob against the new player, Bob loses.

Which points out the biggest problem with strength of schedule:  You now have a legitimate reason for a good player to complain (and possibly just walk out) if they discover that they've been paired with a weaker player.

Flip that around:

Bob wins all three of his games, so does Susan.  Because you paired Bob against the new player, Bob wins.

Which points out the biggest problem with differential:  You now have a legitimate reason for a good player to complain (and possibly just walk out) if they discover that they've been paired with another good player. They know that someone who's been paired with a new player will beat them due to that 10-0 boost to their score.

I'm in favour of anything that doesn't encourage experienced players to smash new players into the ground in order to secure a better position.

[solkan]

 

As opposed to the current system, in which Bob wins because Susan had to play a better player. Bob is being rewarded for not being challenged. You cannot seriously tell me that's a superior option.

Like I said, both versions have a luck aspect, so we have to decide if it's better to reward somebody for smashing noobs, or for slogging through more even match ups. To me, there is no question.

You're not addressing the complaint at all:  That if you use random pairings, Strength of Schedule is just an elaborately determined random tie breaker.

Here's what it breaks down to:

* Strength of Schedule is more work to implement

* Strength of Schedule is still determined by the initial random pairing

* Strength of Schedule doesn't prevent "Strong Player smashes Weak Player"

which leads to a situation where players get punished randomly and arbitrarily for their initial pairing.  It'd be a loss less work to just flip cards as the tie breaker, and flipping cards would probably seem a lot more fair to everyone involved.

 

[solkan]

 

Flip that around:

Bob wins all three of his games, so does Susan.  Because you paired Bob against the new player, Bob wins.

Which points out the biggest problem with differential:  You now have a legitimate reason for a good player to complain (and possibly just walk out) if they discover that they've been paired with another good player. They know that someone who's been paired with a new player will beat them due to that 10-0 boost to their score.

I'm in favour of anything that doesn't encourage experienced players to smash new players into the ground in order to secure a better position.

Differential being bad doesn't make strength of schedule a better alternative.

That's what it comes down to.

 

[Kadeton]

If the choice is between those two, I'll take the one that gives a better experience for new players. If you've got other suggestions, this seems like a great place to talk about them.

[Yipikayey]

If you want to improve the experience for new players, reduce the max differential possible. It won't stop people defeating new players, but they won't require to squeeze them in order to compete.

I said it before, the problem is that a 10 VP differential is really big so you get a big fat check if you happen to face a new player. But if you really want a fair system, you would require swiss pairings from round 1 using a ranking system, and that's not within the realm of possible in most areas.

[Kadeton]

 

If you want to improve the experience for new players, reduce the max differential possible. It won't stop people defeating new players, but they won't require to squeeze them in order to compete.

Do you mean capping the differential, so that (for example) a 10-0 victory and an 8-3 victory would both result in a +5 diff? It's an interesting thought.

I feel like it still advantages those who have more "easy" games over those who have more "hard" games during the tournament, but it certainly would reduce the amount of crushing people feel they need to do.

If I'm honest, I would really like to experiment with a system that promoted close victories over big ones, a kind of reverse differential where the tie-breaker was who had the narrowest margins of victory. There's nothing particularly engaging about a game in which you (or your opponent) clearly don't have a chance to win - at that point, you're just going through the motions. Trying to win by as small a margin as possible would actually be a pretty interesting exercise for the strong players to see just how good their control of the game really was, and it would give the weaker players at least the illusion that they were still in the game til the very end.

[solkan]

To compare the current system and the proposed replacements:

1.  Differential method:  The two players each flip three coins, and compare the number of heads they each flipped.  The person who flipped more coins wins the tie breaker.

2.  Strength of Schedule:  All of the players in the tournament flip three coins.  In order to figure out who gets the bonus point, the tournament organizer enters all of the results into a spreadsheet and hits a button.

From the perspective of the people playing the games at the time, both of those tie breakers are indistinguishable from random chance because they're determined almost entirely by random chance.  Did you randomly get a good opponent?  Then your differential is going to be lower.  Did you randomly get a bad opponent, then your differential is going to be higher.

Same deal with Strength of Schedule.  Were you randomly assigned three above average opponents?  Congratulations, your strength of schedule has been randomly determined to be better.  Were your three opponents below average?  I'm sorry, your strength of schedule has been randomly determined to be worse. 

Do either of those do anything to prevent a strong player from playing against weak players?  No.  Do they both produce opportunities for players to complain?  Yes.

So you're left with:

Differential:

  Pros:

1. The players can do the math
2. The players have some say in the final results

  Cons:

1. Effectively random if opponent selection is random, with better scores the worse the randomly selected opponent is.

Strength of Schedule:

  Pros:

1. If player pairing is not random, then it rewards players for playing against stronger opponents. 

  Cons:

1. If player pairing is random, becomes after the fact justification of results: "We randomly selected this person to play stronger people, so that person should win this tie."
2. Effectively random if opponent selection is random, with worse scores the worse the randomly selected opponent is.
3. More work for the tournament organizer

Note that Gaining Grounds doesn't satisfy the requirements for Strength of Schedule's advantage over Differential, because player pairing is randomly determined in the first round, and the results of the random pairings determines later pairing. 

Flip a card:

  Pros:

1. Honest and transparent about being a random tie breaker
2. Doesn't allow for strong players to get a better score by playing weak players.

  Cons:

      1.  It's flipping a card to break the tie

Objectively, the tie breaker should be replaced with a card flip because it would produce the same result with less work.  Simply because it minimizes the amount of work required to produce the random tie breaker result, and it's not measurably worse than the other alternatives.  It even produces the same opportunities for players to complain about good or bad luck in the tie breaker.

Otherwise, as I can tell people would be better off proposing and developing a system to eliminate the random player pairings completely so the Strength of Schedule tie breaker would actually be an improvement.

 

[monkeyking68]

The only way to make SOS work without complaint would be to have an international rating system for players, similar to what chess does.

[Adran]

I have never played Strength of Schedule, so I am only going by what it seems to mean, but it always worries me. 

I wouldn't not play an event just because of it, but it seems to make the tie breaker seem less fair on the player. I can't find the thread but I can come up with a situation where 1 player in a 12 man event plays 3 games and wins them all 10-0 and doesn't win the event because the person that won their 3 games 2-1 each time had a better "strength of schedule". 

In your example both Buster and Edgar faced Abbey and Fiona They both beat Fiona and lost to Abby. So actually Buster finishing above Edgar isn't all that surprising. After all he did better against both Abbey and Fiona than Edgar did. 

 

[Yipikayey]

 

Do you mean capping the differential, so that (for example) a 10-0 victory and an 8-3 victory would both result in a +5 diff? It's an interesting thought.

I feel like it still advantages those who have more "easy" games over those who have more "hard" games during the tournament, but it certainly would reduce the amount of crushing people feel they need to do.

If I'm honest, I would really like to experiment with a system that promoted close victories over big ones, a kind of reverse differential where the tie-breaker was who had the narrowest margins of victory. There's nothing particularly engaging about a game in which you (or your opponent) clearly don't have a chance to win - at that point, you're just going through the motions. Trying to win by as small a margin as possible would actually be a pretty interesting exercise for the strong players to see just how good their control of the game really was, and it would give the weaker players at least the illusion that they were still in the game til the very end.

Correct. I already had done that in tournaments for infinity, when the system was only a win/loss. Against a newbye, i would maintain the suspension of belief, by purposefully having a close game and make them feel as such, while also teaching them some things, because i wouldn't get punished by it. Once the whole concept of VPs, and different tiers of victorys (the system is similar, top 10 VPs, 5+ VP differential = 3 Points, 1-4 VP differential = 2 Points, this matters a lot), the system gets more competitive, and therefore harder to handicap yourself and still have chances.

If you reward games for how close they are only, while i see that your intention is to force everyone to handicap themselves to be able to accomodate newbies, you are also strangling the competition in a way which is transparently forced, something that won't have a good reception, and i don't think it's fair. I have to admit, that the most interesting games for me, has been sound defeats were i managed to come back to get a decent score and my opponent didn't give me an easy time, since those are what i consider hard fought, not a 5-5 result in a paper. I say this, because hard fought game is hardly something that you can transalte into a result most of the time.

It is true, that if you increase the number of rounds, on a small number of players, it will increase how fair the result will be. And to be honest, that's the only thing we know that really works with swiss pairings, and a rank system. But for regular store tournaments, where there are 8-16 players, on 3 rounds, i think that people shouldn't worry that much about who gets 2nd and who gets 3rd, because the only one who will get a somewhat accurate result are the player who ended first (he won the 3 games), and the who ended last (he lost 3 games). Anything else is somewhat accurate, but kinda flippy, and you can try to change things, but you will keep getting the same level of accuracy because the game sample will still be too low.

 

 

[tomjoad]

 

If you want to improve the experience for new players, reduce the max differential possible. It won't stop people defeating new players, but they won't require to squeeze them in order to compete.

I said it before, the problem is that a 10 VP differential is really big so you get a big fat check if you happen to face a new player. But if you really want a fair system, you would require swiss pairings from round 1 using a ranking system, and that's not within the realm of possible in most areas.

I didn't mention this initially, as I have found people actively hateful of this opinion, but the fact that diff encourages enfranchised players to stomp new players is, in my mind, another strong pro- for using SoS. Diff encourages people to be as competitive as possible and that almost 100% of the time ends with less pleasant interactions, especially for the kinds of people most likely to get beat 10-0. Using SoS means that a win is a win most of the time, so the incentive to beat on and eke out every small advantage over a noob is greatly reduced.

[Yipikayey]

The reason why people can be actively hateful about it, is because it discourages full competition. While i don't play that way in local tournaments, i can understand why undermining the competitiveness can lead to resistance.

[tomjoad]

 

The reason why people can be actively hateful about it, is because it discourages full competition. While i don't play that way in local tournaments, i can understand why undermining the competitiveness can lead to resistance.

My thinking is that is isn't undermining the competitiveness. You still win if you win 8-2 instead. By being a little friendlier you are more likely to grow the community though and even help another player improve.