Seeding by skill

[Supermodel From Paris]

Outline of issues:

- Seeding by skill cannot be accurately or fairly ranked outside a tournament series in which each tournament uses the same set of rules and same environmental factors. While mitigated by the Unity Ruleset, seeding by skill outside of a series still lacks objectivity and attempts to rank players based on virtually nothing but the opinions of a single entity or entities without any concrete or supporting data.
- Seeding by skill undermines the tournament process by attempting to predict a winner or the strongest finishers before it starts. In a tournament environment, only the results should rank the players attending.
- There is no proof that seeding by skill produces more accurate results, only that it produces results closer to those which we expect, which is not the same thing.

I believe instead that we should seed only by location. Specifically, we should seed by zip code or area code. I believe this is a strong seeding method because:

- Seeding by location ensures players who practice together frequently will not be eliminated by one another early in a tournament.
- Seeding by location is truly objective; there is no potential for bias, nor does it attempt to predict tournament results before the tournament actually begins.

It is worth noting that this method is supported by other competitive communities, and has been used for decades within them with great success.

Discuss!

 

[Linkvader]

Does smash seed by skill? If it does, I had no idea...

 

[Supermodel From Paris]

Yes it does.

In other gaming communities, seeding is done almost entirely by area of origin. Every once in a while, a good player attempts to escape their seed by asking TOs to give them a more forgiving opponent in round 1. This usually gets out and their community becomes outraged, because to them, your unfortunate potential "Anti vs. Reflex" situation in round 1 is just the hand you were dealt. No one gets any preferential or special treatment.

To editorialize a bit, I will say that I believe top players complaining about placing badly because they drove far, paid a large amount of money, etc. expecting to place better is nonsense. This line of reasoning is largely responsible for our seeding methods which are designed to cater to players who have grown to expect a certain type of treatment. When you walk in the door, you're just like everyone else until you leave. At that point, you are ranked. Before then, you're just like XXIkeMainXX2000.

 

[san.]

That reminds me when I had to play Ally his first round at MLG.

 

[dainbramage]

It is worth noting that this method is supported by other competitive communities, and has been used for decades within them with great success.

Such as?



Plenty of knockout tournaments (e.g. tennis slams, GSL if you want a game) will seed by skill. Which is subjectively decided by someone who decides who the seeds should be.

Imagine a randomly allocated bracket that had say, m2k, ally, dehf and adhd (lolnamesearchbait) playing each other first round, with the losers playing eachother in the first game of losers bracket, ensuring that one gets knocked out in 2 sets. If you were holding that tournament, would you allow that situation to occur?

 

[Budget Player Cadet_]

That reminds me when I had to play Ally his first round at MLG.

This. Imagine a 256-man tournament bracket, where for some reason, all the pros ended up clustered on one side of the bracket, and on the other side there are pretty much just random scrubs... Say, people I could beat. In this tournament, I could end up in the right bracket and basically just coast my way to third place (losing only winners and losers finals because that's, you know, the first time I actually have to play against a player which is decent), while people like San and Ally are knocking each other out of the bracket early on.

This is just not a good idea. We know that certain members of the community are good, and facing them up against each other right at the beginning of the tournament is a bad idea.

But what about pool seeding? Sure, at some level, it's still "unfair" for the reasons you pointed out (arbitrarily giving those who are established as good an advantage), but it's not nearly as extreme, and the pools results establish the seeding for the rest of the tournament, mitigating a lot of the little hiccups (example: a newcomer who is actually really good could get second in a good pool and then do well in bracket; with direct bracket seeding he'll probably fight some pro round one).

 

[Supermodel From Paris]

Why is that a bad idea, BPC? No one should be allowed preferential treatment. You use the hand you were dealt.

 

[deepseadiva]

Call me crazy

But I see flaws in this.

 

[Elyssa Xey Hexen]

In BPC's extreme example, assuming that the better player always wins, the best two players would still end up getting their proper ranking, yet everyone else would get the shaft.

This is because eventually in tournament, the true 3rd place person would have ended up losing to someone twice anyway. Whether they lost in semi finals or in the first two rounds, it was destined they would have lost and not get a ranking higher than 3rd place. So, for double elimination, tournaments are accurate for the first two places generally. So, BPC is roughly correct, even doing random placement will not work.

Yet, as you said SFP, seeding skews the results to what one would expect which does not seem fair.

Either way, double elimination is not the ideal touranment method for declaring ranking. Round robin is better for that. But, you still have to conform to players needs. People drive/fly up expecting money, they want their money, and seeding helps to fix people's opinions. Also, there are time requirements. Roundrobin takes way to long.

 

[Supermodel From Paris]

So, full disclosure: I was not completely prepared to fight every opposing view point here, and I expected there would be a lot of them. I also suspect not many people will like or agree with my idea. I just thought it warranted discussion. Nonetheless I will try to respond to the points made here.

First, I invite everyone to read this thread: http://shoryuken.com/f2/i-cant-believe-what-i-just-saw-272424/ and read how their reactions are different in this scenario. In Smash, not only does this happen frequently, but even when it does, our seeding method essentially offers the TOs peers and friends this sort of preferential treatment. At every tournament.

That reminds me when I had to play Ally his first round at MLG.

I'm not sure I follow. First, MLG a series of tournaments, not a single tourney. The first tournament in this series essentially served to seed the rest of them. I'm not sure how the first tournament was seeded, so I can't comment on that particular issue.

However, I reject the notion that you or any other player somehow deserves a more forgiving first bracket. In my first tournament in the NYC area no one knew who I was. First round I played Minty, who as at the time, a power ranked Marth player who far outskilled me. In my second round, I played an out-of-state no-name forced to play M2k round 1, but who was actually a really strong player. As a result, I placed dead last. Playing friendlies in the room that day, I can tell you I was certainly not the worst player in attendance or even in the worst tier.

At another tournament a few months later, someone for some reason accidentally seeded me as a "top player". As a result, I placed top 20, which isn't in the money, but is a lot higher. This is because my first round was against a Sheik player who had never played in a tournament, and my subsequent matches were against mid-level players closer to my capabilities. An easy bracket can drastically alter how one places at any skill level.

The problem with your example is that everyone else (players who do not usually place well) deal with this at every tournament, and it is a self-perpetuating cycle. Also worth noting is that you placed 33rd at the first MLG (which I guess is the one you are talking about) with other very good players like Snakeee, Dekar, Kismet, Choice, and Shaky, and out of 183 entrants. Your placement wasn't far off enough that your unfortunate run-in with Ally in round 1 sets off any alarms. NickRiddle also placed 12th, and placed top 5 at subsequent events.

While your match with Ally in round 1 was unfortunate for you, I'd like to point out that someone fights Ally in round 1 at every tournament, and that guy usually loses. I'm sorry it was you once, and I don't see how your status as a top player should give you preferential treatment. Again, in the link above, the SF community is shocked to find out Jwong's seeding was arbitrated to give him an easier bracket, but in Smash, this happens at every tournament.

This is just not a good idea. We know that certain members of the community are good, and facing them up against each other right at the beginning of the tournament is a bad idea.

We also know that statistically this is pretty unlikely. What is more likely in a tournament large enough to have multiple players at the level of Ally or M2k is that one good player is sent to losers round 1 and sweeps through the tournament attaining 13th place or something, which we have seen happen in random seeding scenarios a few times. We have never seen a top player get last because of seeding, though. I doubt the SF community sees it happen very often either. Even if it happened once, I'm not sure that's enough to completely discredit the entire method.

In BPC's extreme example, assuming that the better player always wins, the best two players would still end up getting their proper ranking, yet everyone else would get the shaft.

This is because eventually in tournament, the true 3rd place person would have ended up losing to someone twice anyway.

Gonna stop here to respond to this: this is exactly the problem I have with the current system. We assume we know who the "true 3rd place" player is and attempt to create the bracket to give this player the best possible chance at attaining the place they "deserve" which is extremely broken for a number of moral reasons I shouldn't have to explain. After the next quote I'll explain the other problem with this.

People drive/fly up expecting money, they want their money, and seeding helps to fix people's opinions. Also, there are time requirements. Roundrobin takes way to long.

Maybe they should drive/fly expecting to compete, instead.

Moreover, most of the arguments in this thread so far are already covered in the OP, specifically:

"Seeding by skill undermines the tournament process by attempting to predict a winner or the strongest finishers before it starts. In a tournament environment, only the results should rank the players attending."

and:

"There is no proof that seeding by skill produces more accurate results, only that it produces results closer to those which we expect, which is not the same thing."

If we already know who the best players are, there's no reason to hold a tournament in the first place, especially if we're attempting to fix the brackets to get us closer to the results we want.

Also, I feel that if we are to continue to seed this way, we ought to come up with some kind of ELO system. This is especially possible now with the introduction of a global ruleset. It is only the current method of using the TO's gut to seed that I have a problem with, I guess, not true Elo seeding.

 

[NickRiddle]

I personally hate seeding by skill.
It basically says, "The top players are far apart so they're guaranteed to make it to the top of winner's bracket... Assuming they are as good as their seed suggests."
That, to me, is unfair to "worse" players. They NEVER have a chance of placing high due to the fact that they're going to be playing the same 4/8 people around the same time over and over again.
If you seed by location, those who practice together cannot play early on. That makes more sense to me.

What's worse? Top player A doing poorly because of an early run-in with another top player, or a group of people who drive 8 hours together to your tournament... who have to play each other first round?

 

[ch33s3]

Seeding randomly or by location is accurate to the top 2 (given no upsets). Seeding by skill is accurate to the number of manual seeds (again given no upsets). Smash tournaments reward more than the top 2, and thus not seeding by skill is a disservice to the integrity of the tournament.

Let's assume 3rd gets $1000 and then all the places down to the top 8 also get nice size prizes. If the third best player there is forced to play the best player in winner's and the 2nd best in loser's, they could place in a position that does not pay out due to poor luck and not their skill. On the flip side, someone like the 17th best could trip their way into an easy bracket, and walk away with $1000.

Why the **** would you want to leave thousands of dollars in prize distribution up to random chance?

 

[Supermodel From Paris]

Seeding randomly or by location is accurate to the top 2 (given no upsets). Seeding by skill is accurate to the number of manual seeds (again given no upsets). Smash tournaments reward more than the top 2, and thus not seeding by skill is a disservice to the integrity of the tournament.

Let's assume 3rd gets $1000 and then all the places down to the top 8 also get nice size prizes. If the third best player there is forced to play the best player in winner's and the 2nd best in loser's, they could place in a position that does not pay out due to poor luck and not their skill. On the flip side, someone like the 17th best could trip their way into an easy bracket, and walk away with $1000.

Why the **** would you want to leave thousands of dollars in prize distribution up to random chance?

I invite you to investigate this issue in the fighting game community and find one incident in which a bad player walked away with $1000, or even a totally surprising underdog that was never again able to repeat their victory.

In any event, I prefer Elo to "by location" seeding, but since we don't have any effort to produce an Elo seed, our current method of "skill" seeding has more problems than location seeding.

In addition, you are using a strange definition of accurate, something I covered in the original post. It is accurate given what we expect, but there is no proof what we expect is more correct than results from location-based seeding or even random seeding. If anything, there is a lot of evidence from other parts of the FGC that suggests the total opposite. Your desire to make sure the "correct" players are placing correctly while in line with the status quo, is actually pretty non-competitive and undermines the tournament process.

I personally hate seeding by skill.
It basically says, "The top players are far apart so they're guaranteed to make it to the top of winner's bracket... Assuming they are as good as their seed suggests."
That, to me, is unfair to "worse" players. They NEVER have a chance of placing high due to the fact that they're going to be playing the same 4/8 people around the same time over and over again.
If you seed by location, those who practice together cannot play early on. That makes more sense to me.

What's worse? Top player A doing poorly because of an early run-in with another top player, or a group of people who drive 8 hours together to your tournament... who have to play each other first round?

Funny you should say this. I was actually seeded against Minty at like 3 consecutive tournaments, lol.

 

[Tarmogoyf]

Also, as an example, competitive MtG does something like this. At Grand Prixs, they award byes (it's swiss format) based upon your rating (although they have national rankings, so it's more objective), so that the best players don't have to play each other early in the tourney.

It's perfectly fine if you seed it by location as well as skill. At small locals in amazing regions, sure you might get a problem, but it's totally possible to do at any OoS tourney lol.

 

[ch33s3]

The MLG events had at least partially random seeding, and you can see how wonky those results are in many cases, imagine if they'd been entirely random (the first event was seeded only slightly, the remainder were done by previous points and anyone who had 0 was randomly seeded below everyone with points).

Edit @ nickriddle: if lesser players lose to the same 4/8 people, why should they ever outplace them? That's like me losing to ADHD every week (which I do), but sometimes making more money than him. Why does that make sense?

 

[Supermodel From Paris]

The MLG events had at least partially random seeding, and you can see how wonky those results are in many cases, imagine if they'd been entirely random (the first event was seeded only slightly, the remainder were done by previous points and anyone who had 0 was randomly seeded below everyone with points).

Edit @ nickriddle: if lesser players lose to the same 4/8 people, why should they ever outplace them? That's like me losing to ADHD every week (which I do), but sometimes making more money than him. Why does that make sense?

I'll say it one more time, and then I'm not responding to this point anymore or we will start going in circles:

There is no proof that seeding by skill produces more accurate results, only that it produces results closer to those which we expect, which is not the same thing.

 

[ch33s3]

Just to be clear here, do you think:
1. Ally
2. Anti
3. ADHD
4. SFP
Is more accurate than:
1. Ally
2. Anti
3. SFP
4. ADHD
?

Assume you placed third because you played nonames through winners and lost winners and losers finals.

If the answer to the above question isn't yes, you're delusional.

 

[Supermodel From Paris]

I do, but I also don't think I would ever place that high even in a situation where I was handed a relatively easy bracket. Remember, the people I'm playing through Winners were also, surprise, winners.

The best I could hope for in this situation is like a 17th place finish like I got when I first moved to New York, but good players are given pretty artificially easy brackets a lot, too.

Again, I'd love it if you would research fighting game history and find a situation in which their method caused this problem.

 

[san.]

My comment wasn't an argument for any side, by the way. I just said it reminded me of it, that's all. (and it was seeded for like the top top players I just had a random bottom seed).


I don't really have much of an opinion. I can't see the difference between seeding by skill and seeding by location other than personal preference.
In other words, I can't think of an argument that favors location seeding but disproves skill seeding.

 

[ch33s3]

Using you as an example was just convenient. I could have used myself to be a tad more realistic.

And you're trying to change the status quo, so it's on you to show how seeding by location is as accurate (as per a definition we all agree on, I suggest all players that get money deserve it and nobody who does deserve it does not, based on their skill level compared to others present. Upsets are great, just assume the lower skilled player is as good as the best player they beat.)

Off the top of my head though, since I don't closely follow fighters, I can only think of a few examples quickly. Diem and zucco, particularly zucco, having very rough early matches comes to mind because I've seen them complain. I've also seen some finals class matches in the first couple of rounds on WNF. If you really see the need for me to find someone who got screwed out of cash by brackets because of non-skill seeding I'm sure I can when I'm not mobile, and it happening even once is too many times.

 

[ADHD]

My comment wasn't an argument for any side, by the way. I just said it reminded me of it, that's all. (and it was seeded for like the top top players I just had a random bottom seed).


I don't really have much of an opinion. I can't see the difference between seeding by skill and seeding by location other than personal preference.
In other words, I can't think of an argument that favors location seeding but disproves skill seeding.

Hi san.

10:D

 

[Mic_128]

If you really see the need for me to find someone who got screwed out of cash by brackets because of non-skill seeding I'm sure I can

Pretty sure he asked for a time a 'bad' won money, not when someone didn't win money because of brackets.

 

[MK26]

sfp, the problem with your situation of getting a death bracket in your first tournament came about from a limitation inherent in all tournament seedings - the inability to accurately rank an unknown. You could be good, you could be great, you could be average, you could be ****, but if you have no previous rankings, then the the TO has no choice but to assume you're not good. seeding by skill is a rewuirement because you cant make assumptions other than that previous ranking will continue the way they were. As for why the best player is seeded against the worst, look at it this way:

In a 32-man tourney, 1st seed plays 32nd seed in the first round. Given an ideal bracket, 1st will go on to play 16th, 8th, 4th, 2nd, and 2nd again in GF's before winning the tourney. That is the assumed outcome based on past results, and if there are no upsets then it will be the most 'correct' result. You might say 'that's unfair, he got to play 5 people seeded lower than him!' but do you know why he got to play 5 people seeded lower than him? Because hes the best. There is nobody above him for him to play. By virtue of having the highest skill level, everybody he plays will automatically be below him.

Now, imagine that, for some reason, the 32nd ranked seed pulls off a stunning first-round upset and wins. Now, look at the players he has to face! 16th, and if he wins again 8th, 4th, 2nd, and maybe even 1st if he comes back from losers'! that's not at all fair! I mean, hes playing everybody above him! But wait just one second. He's ranked 32nd, so theres nobody below him for him to play. Furthermore, beyond the first round match, the first-ranked player and the last-ranked player have the exact same people to go through. Read that again. The exact same people. The perception of unfairness comes from, like i said, all of these players are below the best player (so it looks like he has easy matches) and above the worst player (so it looks like he has hard matches), where the matches really are all the same and its the skill of the first- or last-place player that actually varies.

 

[Flayl]

http://www.smashboards.com/showthread.php?t=293399

I don't know who actually won money

 

[Zankoku]

This is actually intriguing to me, so I decided to run some experiments, with a 64-person bracket. I can't access my web server from here, so I won't be able to upload my notes until later tonight, but from my initial testing I've determined that, with top 6 payouts, random seeding and "soft skill" seeding (where the TO knows the top 16, the bottom 16, and the rest, but not the order within each of those tiers) have roughly equivalent accuracy. Placements outside of the money had a decently higher degree of accuracy with skill seeding, but that's to be expected.

Purely random seeding when running pools, however, leads to potentially disastrous results.

 

[Supermodel From Paris]

Cheese, I already checked to see if here were enough examples to cause concern. If you really believe you they're there, it's up to you to find them, not me. I'm not going to make your argument for you.

Ankoku, thanks. With that said, I now certainly prefer this method over "soft skill" seeding because while expected players are out early occasionally, players who win money are mostly within the realm of reason and we would see a lower amount of bias and TO/peer collusion which is pretty rampant (accuasions of TOs giving themselves easier brackets than expdected, etc.) The difference, at least at the mid level of play, between a last place finish and a 9th place finish seems mostly to be how the TO ranks their skill level at the moment, anyway. However, I don't like the way you used "accuracy" in your post because it's too assumptive.

In any case, I would like to re-iterate my firm belief that a true global rank-based Elo measurement is superior to either method as it removes TOs and TO "assistants" from the equation entirely and ensures the maximum fairness. Maybe this is something I can research and do a little work on myself.

 

[Zankoku]

I don't know what other word I can use besides accuracy. The only way to isolate how well the methods work without outside factors is to assign each "player" to a definite level (a lower number will always defeat a higher number), in which case accuracy is how close to being in proper ordering the end results are for the numbers.

When you get 25 and 30 placing dead last and 47 making it into the top half, it's obvious that completely random seeding will have an effect on pretty much anyone who isn't actually in it to win it. Money placement wise, random isn't harmful, but assuming you seed based on Elo, which is based on matchup results, you might be perpetuating inaccurate seeding by starting with complete random.

Here's a quick excerpt from my notes:

AVERAGE RANDOM
1 $256.00
2 $160.00
3 $64.00
4 $57.60
5 $44.80
6 $32.00
7 $12.80
11 $12.80
8 $0.00
9 $0.00
10 $0.00
12 $0.00

AVERAGE SOFT SKILL
1 $256.00
2 $160.00
3 $96.00
5 $53.33
4 $42.67
6 $10.67
7 $10.67
9 $10.67
8 $0.00
10 $0.00
11 $0.00
12 $0.00

I suppose the most significant factor was that the top 3 consistently placed in the top 3 with "soft skill" seeding, since it ensured that none of the top 16 would ever have to play each other early on and thus made 3rd place much less likely to run into 2nd or 1st too quickly.

 

[ch33s3]

SFP, I of course agree that some kind of unified ranking would be the best. It'd be very stale matches played over and over but it's most fair.

I provided you with a whole season of MLG results to pull up. Those are semi-random and look at how many times "lesser" players made money, or top players did not. Felix getting third, with absolutely no offense meant to him, is certainly a huge anomaly. MLG was not random, merely more random than solid skill seeding. If things like that happen when the top players are seeded, multiply the effect for fully random bracket.

Ankoku, if I'm interpreting your post right, you "paid" top 8. That means 1-8, since we're assuming no upsets, should win 100% of the money, and nobody should ever win more than their portion of the pot based on seed. The fact that both of these happen without upsets in both random and soft skill seeding should make it apparent that anything but hard skill seeding is unacceptable.

Unless you like randoms winning money at smash tournies of course.

 

[Zankoku]

I paid top 6, not top 8. A single instance of 7 winning money and a single instance of 11 winning money isn't exactly something I'd consider "unacceptable," since in the real world it's pretty damn hard to rank people exactly where they belong.

EDIT: I also took a look at the results that you were referencing. Felix beat MikeHAZE, you, and ESAM in brackets before finally losing to M2K and LeeMartin. Whether or not he is a "lesser player" who doesn't deserve third place, this is a case of someone pulling at least two major upsets in brackets and beating higher seeded players, not lucking out on random seeding and having an easy ride to finals.

 

[Player-1]

The whole entire point to a tournament is to define who is the better/best player. If the tournament isn't doing that then it's not doing it's job, if you have the 3rd best person getting last place because they had to play 1st best then 2nd best person back to back in winners and losers then the tournament isn't doing it's job. That's why seeding by skill happens, the fact that other fighting game communities don't do that is completely ********

 

[Zankoku]

The whole entire point to a tournament is to define who is the better/best player. If the tournament isn't doing that then it's not doing it's job, if you have the 3rd best person getting last place because they had to play 1st best then 2nd best person back to back in winners and losers then the tournament isn't doing it's job. That's why seeding by skill happens, the fact that other fighting game communities don't do that is completely ********

How can that even happen? This implies that the second best player lost to someone who wasn't the best player, since getting last while losing back to back to first and then second is only possible if you lose to #1 in round one.

It'd be something like this:

Note that you're already getting the extremely unlikely, even in random seeding, event of four top players right next to each other in brackets. Next note that 2 has to lose to 4 in order for 3 to encounter 2 in losers' brackets.

If you're trying to make the argument of 1 beating 2 in the first round, then 3 in the second round, and 3 getting ****ed because of that, you'll note that due to bracket design, 3 actually will not face 2 until losers' semifinals, guaranteeing himself at least a 4th place spot if he doesn't lose before then.

 

[ch33s3]

I paid top 6, not top 8. A single instance of 7 winning money and a single instance of 11 winning money isn't exactly something I'd consider "unacceptable," since in the real world it's pretty damn hard to rank people exactly where they belong.

If you don't find that unacceptable, we're diametrically opposed and I'm not discussing this further.

 

[xDD-Master]

Pools that are seeded -> pool-results determine skill for bracket -> bracket

Best method. Worse Players can do better then expected in pools *upsets* while the good players will very likely still be first in their pools so they don't have to play early in Bracket. They still can get second/third, which isn't THAAAT bad.
Pools+Bracket >>>>>>>>>> Only Bracket

A tournament with more than 32 players should ALWAYS have pools.
Tournaments with 16-32 could depend on the avarage level and/or how many top players are there.
Then Skill seeding into Low/Mid/High would still be better then completetly random.

And Location should ALWAYS be kept in mind when seeding.
By saying that, a hybrid of Location+Skill is THE best.

So all together:
Pools with Seeding+Location -> Pool results determine new Skill Level -> Bracket with Seeding+Location (+ Keeping an eye, that no players from the same pool play each other too soon again)

 

[NO-IDea]

I find this topic interesting.

Cheese, I don't think you should undermine Ankoku's analysis of his results. Let's be honest. I can't think of one example where a player who placed 9th-12th at a national did not have the capability into placing into top 8 (quite literally they were 1-2 victories away from that placement anyway! The skill differential dramatically decreases as you go past top 5/8/16 (local/regional/national).)

I don't find how this is unacceptable.

 

[ch33s3]

I find this topic interesting.

Cheese, I don't think you should undermine Ankoku's analysis of his results. Let's be honest. I can't think of one example where a player who placed 9th-12th at a national did not have the capability into placing into top 8 (quite literally they were 1-2 victories away from that placement anyway! The skill differential dramatically decreases as you go past top 5/8/16 (local/regional/national).)

I don't find how this is unacceptable.

Of course, any player on any day can upset a better player. That's not in dispute here.

However, we can't control for upsets in any meaningful way. You can only assume no upsets will occur when seeding, and that is the only fair way to create a bracket and get the most meaningful results. Sure if Seed 32 beats Seed 1 and places really well the bracket will be a little skewy, but next tourney Seed 32 certainly won't be Seed 32 anymore, and the results will again stabilize.

EDIT: I think that pools (again seeded by skill and loosely by location) -> Amateur and Pro (top from AM advance to Pro) brackets is the best way to do things, but there is NO way to not use skill seeding and have anything close to a fair bracket.

 

[Zankoku]

lol.

Players can't be as solidly classified as numbers. You see two isolated instances of each of the seeding methods of people just outside of the top 6 placing in the top 6 and find that unacceptable, but you're fine with a seed 32 suddenly performing better than he should, causing an "upset," and then disregarding this seeding error because "results will again stabilize"? This is wholly hypocritical - you're saying seeding in a non-exact method because it's impossible to truly account for the absolute skill of each player is unacceptable, but seeding as much by the numbers is possible even if the first few tournaments are gonna end up with upsets until everything finally reaches the proper number... and how many majors are you willing to sacrifice to get to this "stability", assuming everybody else stays in the same relative position (god knows that 12th seed might suddenly come to an epiphany and become worthy of 5th seed, or that 9th seed might have a few good matchups against 7th and 4th and randomly place 3rd at one tournament)?

I understand the desire to maintain as much accuracy as possible, but when a single mistake results in placements that are completely off, then you're sacrificing your possible margin of error in exchange for claiming that you know where each and every person belongs.

 

[NO-IDea]

Cheese: When you look at results, do you really care the order of who was placed beyond 5/8/16 (again depending on size)? I don't.

According to Ankoku's results, the players who were supposed to place in the money did. Random seeding didn't ensure accuracy beyond certain placements, but who cares at that point?

To take this into perspective, in a local, aren't the money-placers usually the same top 3-5 every time? Even if you were to random seed, I would have full faith in their skill to seize all the money-placing spots before a lesser skilled player could.

If upsets occur and an unexpected player takes a money spot, then maybe the "lesser skilled" player isn't as bad as you thought. Case in point, Ankoku proved earlier how Felix deserved his spot. As far as I'm concerned, even in my own region, if I were to devise a random seeded tournament, I'm 100% confident that our usual top five players will place in top 5. The order of which they place I cannot determine, simply because the differential between their skill is so small.

Move this into a larger spectrum, at a regional you can expand this to the expected top 8. Even through random seeding, as an outsider you can predict who's still going to place top 8 regardless of initial seeding. This in fact is because of double elimination. I would agree that if this were single elimination, random seeding can have devastating results. But the system already works out the kinks.

And moving on, top 16 at a national is similar. You, as a player, know who's the best and who's going to place. And if a usual 17th or 33rd placing strangely takes a spot in top 16, well he probably caused some massive upset and deserved his spot.

I don't see what's wrong with random seeding as long as those that should place in the money will. We shouldn't care what occurs after that.

I don't want to get into pools. The general consensus already is that pools will cause the most accurate results and the only reason why it isn't utilized is because of the organization, equipment and time necessary to make it effective. No reason to argue it further.

 

[ch33s3]

lol.

Players can't be as solidly classified as numbers. You see two isolated instances of each of the seeding methods of people just outside of the top 6 placing in the top 6 and find that unacceptable, but you're fine with a seed 32 suddenly performing better than he should, causing an "upset," and then disregarding this seeding error because "results will again stabilize"? This is wholly hypocritical - you're saying seeding in a non-exact method because it's impossible to truly account for the absolute skill of each player is unacceptable, but seeding as much by the numbers is possible even if the first few tournaments are gonna end up with upsets until everything finally reaches the proper number... and how many majors are you willing to sacrifice to get to this "stability", assuming everybody else stays in the same relative position (god knows that 12th seed might suddenly come to an epiphany and become worthy of 5th seed, or that 9th seed might have a few good matchups against 7th and 4th and randomly place 3rd at one tournament)?

I understand the desire to maintain as much accuracy as possible, but when a single mistake results in placements that are completely off, then you're sacrificing your possible margin of error in exchange for claiming that you know where each and every person belongs.

I think you misinterpreted my point. When seeding one cannot, absolutely cannot, account for upsets, better matchups, or anything like that. You must go with the purest seeding criteria possible (in a perfect world, pools, but in the real world, power rankings). If someone gets an upset great for them, but you can't worry about that in seeding world. All an upset means is that a lower seed will place one spot place higher than they "should" and the higher seed will place one place lower than they "should", which is actually very unfortunate, but it's not set in stone, it's real life, so seeds can change over time. If we were perfect at seeding, the same matches would happen every time and it would be perfectly accurate (one person from each group, say the 4 9th placers, would be ordered within their pack, everyone below them would lose to each of them, and none of them could beat any of the 7th place finishers). There are too many variables to be perfect every time, but any variables we can control, we must. Random seeding just adds variance to the tournament, and that is never good for a statistical analysis of a group (which is what a tournament amounts to).

Cheese: When you look at results, do you really care the order of who was placed beyond 5/8/16 (again depending on size)? I don't.

According to Ankoku's results, the players who were supposed to place in the money did. Random seeding didn't ensure accuracy beyond certain placements, but who cares at that point?

To take this into perspective, in a local, aren't the money-placers usually the same top 3-5 every time? Even if you were to random seed, I would have full faith in their skill to seize all the money-placing spots before a lesser skilled player could.

If upsets occur and an unexpected player takes a money spot, then maybe the "lesser skilled" player isn't as bad as you thought. Case in point, Ankoku proved earlier how Felix deserved his spot. As far as I'm concerned, even in my own region, if I were to devise a random seeded tournament, I'm 100% confident that our usual top five players will place in top 5. The order of which they place I cannot determine, simply because the differential between their skill is so small.

I don't see what's wrong with random seeding as long as those that should place in the money will. We shouldn't care what occurs after that.

I don't want to get into pools. The general consensus already is that pools will cause the most accurate results and the only reason why it isn't utilized is because of the organization, equipment and time necessary to make it effective. No reason to argue it further.

I care about the full placement order of every tournament, of course. Anyone interested in real results should, it's just as important a part of the sample as any other, first, ninth, or ninety-seventh. Ankoku's analysis had at least two players who "didn't deserve" (weren't upset, did not cause any upsets) that placed in the money.

I'm not at all confident that a random seeded tournament in MD/VA, NY/NJ, or any other region with a decent amount of skilled players would be remotely accurate.

The monetary prize is not everything for everyone. A high placement is, and should be, a prize to any player who is passionate about the game, whether a cash prize accompanies it or not.

And yes, we all know pools are awesome, and the logistics behind them are unfortunate.

 

[xDD-Master]

Placing in the money isnt all lol
There is nothing wrong with seeding.
Like cheese said, placing themselves is also important.

Every good sports-community does it (If we count e-Sports as sport). Or at least they should.
Everyone who does completety random pools+bracket are bad.

It's not like no one knows that Ally/M2K and all the like are the best.
If they play bad in pools, for whatever reason, they can still get a hard bracket.
But it's not acceptable when the best players have to fight each other in the first round.

It's just bad. The only reason you're able to critizize it, is because we play Double Elimination.
Would you still want no seeding even for Single Elimination tournaments?
I hope you won't say yes o_O

I somehow can agree with the idea of "resetting skill" in a new tournament series, just like MLG did.
But if we see all of our tournaments as a series with near to the same attendance it's legit to use seeding.
And like I already said, pools should ALWAYS happen.
Since in pools you play many oppenants so seeding mistakes will be solved anyway by it's own, and then these pool results should be the new seeding.
(1st Place High, 2nd Mid, 3rd Low, 4th Bottom for example).

You placed a random noob lowest seed and you realize it was a mistake since he now beats everyone?
No problem, pools will show, if he really is soooo good, he will place at least 2 anway in his pool.
But guess what, in bracket only, he would have faced one of the bests, it's either he wouldve lost: "Yeah we predicted it our seeding was gooood" or "DAMN we shouldve seeded him better, but how should we know >___<"
=> reason why pools ALWAYS are better and should always happen if time is there.

If not enough time is available a rough seeding is best: Top/Mid/Bottom or something like that.
Random only is good if you really don't know any players and/or their skill level.
Or if it's to hard to determine who is better and who not.

Lets say 9B, M2K, Ally & Glutonny in a 4 player Bracket. Seeding is not needed at all and not even possible. But it would be good to at leats seperate M2K & Ally cause they play VERY often.

EDIT: I think that pools (again seeded by skill and loosely by location) -> Amateur and Pro (top from AM advance to Pro) brackets is the best way to do things, but there is NO way to not use skill seeding and have anything close to a fair bracket.

You mean like let's say.
80 people tournament.
8 * 10 per Pool.
Top 3 advances.
All that placed 4th play in a bracket for last 2 bracket places?
=> 32 Bracket

Or what do you mean :o

 

[Zankoku]

I think you misinterpreted my point. When seeding one cannot, absolutely cannot, account for upsets, better matchups, or anything like that. You must go with the purest seeding criteria possible (in a perfect world, pools, but in the real world, power rankings). If someone gets an upset great for them, but you can't worry about that in seeding world. All an upset means is that a lower seed will place one spot place higher than they "should" and the higher seed will place one place lower than they "should", which is actually very unfortunate, but it's not set in stone, it's real life, so seeds can change over time. If we were perfect at seeding, the same matches would happen every time and it would be perfectly accurate (one person from each group, say the 4 9th placers, would be ordered within their pack, everyone below them would lose to each of them, and none of them could beat any of the 7th place finishers). There are too many variables to be perfect every time, but any variables we can control, we must. Random seeding just adds variance to the tournament, and that is never good for a statistical analysis of a group (which is what a tournament amounts to).

Again, the more exacting you attempt to be with what is definitively limited information, the less room you have for actually being wrong with a seed number. If you somehow had five people with identical ELO ratings, would you then try to assign each of them in order based on, what, location? "This guy's from Oregon, so he's obviously the worst even though his ELO's the same" or something? I'm all for aiming for optimality, but not when it comes at a definite expense of practicality. I mean, by far the most accurate way to determine results would be to have everyone play in a single round robin. But it's obvious why we don't do that for tournaments with more than eight people.

Let me know when you can tell without a doubt how good everyone is in absolute terms in a region, let alone a country.

I care about the full placement order of every tournament, of course. Anyone interested in real results should, it's just as important a part of the sample as any other, first, ninth, or ninety-seventh. Ankoku's analysis had at least two players who "didn't deserve" (weren't upset, did not cause any upsets) that placed in the money.

I'm not at all confident that a random seeded tournament in MD/VA, NY/NJ, or any other region with a decent amount of skilled players would be remotely accurate.

The monetary prize is not everything for everyone. A high placement is, and should be, a prize to any player who is passionate about the game, whether a cash prize accompanies it or not.

And yes, we all know pools are awesome, and the logistics behind them are unfortunate.

Given real-world mechanics and statistical significance, I can't see how two minor anomalies (oh ****, 7th seed placed 5th this time!) suddenly completely invalidates an approach. If anything, there would be a better argument for the complete mess of results in the bottom half due to zero seeding. However, you can't even perceive skill levels in such absolute terms normally, anyway! Double elimination brackets give tons of ties, and unless you then ask each of those people to go and play tiebreakers, even seeding based on past results is going to give you uncertain numbers. Seeding based on matchup history (ELO)? Still uncertain for quite a while, since a top 4 level player might've gotten unlucky one day and gone 0-2 in brackets, hurting him for future tournaments as you proceed to seed him low enough to be guaranteed first round losers' the next time. By the way, I could've easily tripped you up by "gaming" the seeds and seeding only the top 6, rather than the top 16, into tier 1 of 3, and I would've had 100% consistency with everyone who deserved money placing in the money... Note that seeding the top works best if you seed it with X people, where X is the number of people you know for sure belong up top.

When setting up a tournament, it's important to know two things:
1. you don't actually know everything, and
2. **** happens

That upset of 3rd seed losing to 9th seed might not have been a true upset, but because you just went and made an assumption to place one over the other, an entire section of the bracket is suddenly compromised. Obviously, if you have a year of results and the same attendants every time, there's no reason not to seed everything exactly, but when your series of locals ends with a regional and you get all sorts of out-of-state attendance, you need to realize you don't know exactly where everyone belongs. You can't just fix this for next time, because it's the series finale - there is no next time. Clearly, the best course of action is to seed as best you can and allow for the phenomenon of a number of players all being somewhere in that top level of play, without giving favor toward any one of them because, objectively, you can't.