UPSB v3
Pen Spinning Relations / [topic][4.11.1] Feasible seeding tournament using pooling
-
Date: Thu, Mar 26 2009 03:24:14
This is a "feasible" (meaning requiring minimal amount of organization) tournament with local seeding approximation.
Because the seeding in this case is approximate, using a pool system will alleviate the approximation error. It is meant to filter a large amount of first round videos in a very "quick" fashion to reach the "end" of the tournament faster.
Here's the basic idea:
1. Each community submits a list with an equal number of participants (i.e. 8). Each list must be ordered by strength. In other word, each community is in charge of ranking their own participants in order. I believe this is a reasonable demand which is fulfillable by communities on their own. We refer to the spinners ranked #1 by their own communities as 1-seeded spinner, #2 as 2-seeded, etc.
2. We will create a number of pools which is a multiple of the desired number of spinners in the second round. For instance, if we want 16 spinners after the first round, we can create 1, 2, 4, 8 or 16 pools. For each pool we will take the top Y spinners, where X * Y = number of desired spinners, X = number of pools. e.g. 4 pools, take top 4 of each = 16 spinners.
3. We randomly distribute the 1 seeds in each pool, then the 2 seeds, then the 3 seeds, etc...
4. Judging must select the top Y vids of each pool. This is the most difficult part of this system, because they are not 1vs1 battles. However, I believe that it's feasible because this system is ONLY used for the first round, where the quality standard is not very high. That alone should allow the judges to dismiss a fair amount of the videos of each pool. Then there are probably some outstanding videos that deserve to get in. If there are more oustanding videos than positions available, then the judge needs to go into more details considering those oustanding videos only. If there are less oustanding videos than positions available, then all the oustanding videos are in, and then the judges must carefully consider between a small subset of the "next best" videos. I believe that by proceeding carefully like this, you will only consider a few videos at a time only.
5. When judging, the judges must not only decide who the top Y is, but rank the vids in the top Y. This means there is another implicit "subround" of judging where the judge must carefully look at the top Y vids of each pool and rank them accordingly. If Y is chosen to be small enough, this is still manageable.
6. The end results is that we get ranking of the top Y of each pool. Again, we call the top spinners of each pool the 1-seeded spinners, then 2-seeded, etc.
7. We can now draw a tournament tree for the rest of the tournament. Place all 1-seeded spinners at the "outer" edges of the tree, each facing a random lowest-seeded spinner. other battles are the "inner" branches
8. Follow the tree until the final.
-> The idea here is that because pools can be rather large, on average, the skill level in each pool evens out. This means it won't matter in which pool you are, it will be equally difficult to qualify for next round. This also means that the seeding coming out of those pools is meaningful, because your video was compared against a lot of other videos.
-> The flaw here is that seeds are not equal inter-community. The 1-seed of one community could be weaker than the 8-seed of another. This is why using pools *should* even out the groups. Of course, this hasn't been tested properly yet.
EXTENSION: To alleviate the above problem, the organizers of the tournament could rank the communities by strength. This is a bit difficult to do, but at the least the general idea (seperating strong boards from newcomers for example) is already good.
The idea is then that when we place seeds of the same level in a pool, we must also make sure that each pool contain a balanced mix of seeds from strong and weak communities.
-- Think about each community submitting 8 spinners for 4 pools, top 4 spinners each. If we have 128 spinners (large upper bound), we would have 32 spinners in each pool, top 4 advance. With this number of spinners, the law of average should work out well IMO. The other thing is that spinners won't be able to "relax" on the first round, so you can expect quality videos right from round 1. Furthermore, with only 16 left after that, the tournament will be shorter (5 rounds). Plus how they do in the first round determine their seed for the rest of the tournament, which means it's very important to have a good round 1 video. -
Date: Thu, Mar 26 2009 15:42:08
This sounds like a really fantastic system!
For the WT next time around? I don't know of any other tournaments that are big enough to test this out on before then, unless someone organised an inter-board tournament of some sort inbetween? -
Date: Thu, Mar 26 2009 16:20:02
I like it. However, I would suggest that there would be a seperate pool for say, the 2 next best spinners per round, like a wild card thing.
-
Date: Thu, Mar 26 2009 18:48:35QUOTE (hoiboy @ Mar 26 2009, 12:20 PM) <{POST_SNAPBACK}>I like it. However, I would suggest that there would be a seperate pool for say, the 2 next best spinners per round, like a wild card thing.
i guess, i think that little details like this depends on the circumstances i would do a case-by-case analysis of a tournament and make a custom structure for that tournament. the big picture is more important here.
thats because a lot depend on the ressources available for the tournanet: number of participants, number of judges, time, communication (how good people are at understanding english) -
Date: Fri, Mar 27 2009 02:04:09
Hm, why not rank boards too? That would help with some 8 seed being better than the 1 seed from another community.
We can use rankings from the World Cup. -
Date: Fri, Mar 27 2009 02:17:36QUOTE (Charlie @ Mar 26 2009, 10:04 PM) <{POST_SNAPBACK}>Hm, why not rank boards too? That would help with some 8 seed being better than the 1 seed from another community.
We can use rankings from the World Cup.QUOTEEXTENSION: To alleviate the above problem, the organizers of the tournament could rank the communities by strength. This is a bit difficult to do, but at the least the general idea (seperating strong boards from newcomers for example) is already good. -
Date: Fri, Mar 27 2009 02:33:52
haha, I guess I missed that. Well, do world cup ranking's seem fair?
-
Date: Fri, Mar 27 2009 02:44:53
not really
swespin is #2 yet they're almost dead right now
it has to be something more flexible -
Date: Fri, Mar 27 2009 02:48:38QUOTE (Zombo @ Mar 26 2009, 06:44 PM) <{POST_SNAPBACK}>not really
swespin is #2 yet they're almost dead right now
it has to be something more flexible
Hm, good point. There's always the option of a poll. Not too sure really. -
Date: Fri, Mar 27 2009 03:33:42
i would rather keep the ranking private.
if pool allocation becomes deterministic because of it, then i don't want communities faking their ranking list to gain any sort of advantages. i want the communities to honestly rank their own spinners correctly. -
Date: Fri, Mar 27 2009 03:39:44
Rawr, I thought about this before, like a Board Ladder, where the boards send lineups and battle each other, in a system like the UPSB ladder.
-
Date: Fri, Mar 27 2009 16:32:20
ya ladder is good, i was thinking like:
after WC10, we will have seeding of top 4 teams or so. in fact even now we have the seeding of the top 4 teams (jeb, swespin, ppp, gpc)
then at anytime, a community can arrage a battle against one of the top 4 teams.
if the top 4 team lose, the other team takes his spot in the top 4 and the other go in the other spot.
like if its #1 vs #4, then if #1 lose it becomes #4 and #4 becomes #1.
and then if #2 fights an unseeded team, if #2 lose than it becomes unseeded, and the unseeded becomes #2. the old #2 becomes unseeded.
then at the start of the next WC, we use those seeds to place the teams in pools.
there would have to be some rules on how many challenges a team can accept or refuse as well
but this is something totally different from this topic, which is more meant for individual battles, not team.
this would tak ea lot of work to arrange as well, so its best toonly tihnk of tournaments for now. -
Date: Fri, Mar 27 2009 19:19:23
Mm, sounds interesting. Also sounds like there will be a lot more interaction between boards. A much more active ps world sounds nice.
-
Date: Sat, Mar 28 2009 02:38:00
Hm, so basically, communities can battle each other during the year to get a seeded rank for WC or WT or whatever? Sounds good, and it will make the boards interact more.
-
Date: Sat, Mar 28 2009 11:57:21
And with interaction, brings improvement to those who really want it...Leading to great opportunities.
-
Date: Mon, Mar 30 2009 06:55:20
Great idea. I was thinking of posting the exact same idea prior to your post - you beat me to it.
-
Date: Mon, Apr 6 2009 15:35:24
I think this idea is really excellent, but I just want to add something : In my opinion, it will be better if a description of how to rank spinners in a board is added. The most ranked pen spinner in a board should not be the most appreciated one, but the one who could have the better result...
-
Date: Tue, Apr 7 2009 00:53:51
What I've been thinking:
What I've collected so far: (correct me if I'm wrong).
1. Boards are ranked relative to each other. The seeding system and the Board ladder will affect the rankings.
2. Spinners are ranked by the board itself. Each board might have a unique system, but the important thing is that the spinners are also ranked themselves.
3. Stronger boards are weighted, so that the strongest spinner from the best board is ranked above the strongest spinner from the "worst" board.
So for WT and whatnot, the top 64 ranked people are given spots, seeded 1-64. That eliminates the random pairing, which ensures that #1 and #2 don't face off in round 1. (Brackets are like single elimination).
And for WC, boards are seeded between pools. The team selected by the board varies from board to board.
So yeah... any comments on this system?