A URC members thoughts on the Metaknight Ban

[Dr. Tuen]

Surveys of top 8 are flawed, the top 32 of a national is much more relevant than the top 4 of a solo city tournament.

That's true if you only consider "over-centralization" to be a national issue. I believe the community thinks otherwise, and that a broader investigation would be of more use in perusing that line of thought.

 

[Ussi]

John has a average money made per use and per player i think. least i think i saw one before.

 

[Dr. Tuen]

It's the best thing we have and it is a hell of a lot more accurate a determining character dominance than money will ever be. again, Would we ban Snake for winning a one million dollar tournament?

In hypothetical terms, when utilizing the monetary system of measurement, we might.

However, this assumes a few things.

1. The analysis of 'overpowered-ness' is only facilitated by the monetary metric

2. The monetary metric is the only factor considered when instating a ban (i.e. over-centralization is not considered).

 

[Flayl]

Tuen: I assumed over-centralization is something you look for in the top of a metagame, not in places where the skill gaps can vary significantly between ranks.

 

[Dr. Tuen]

Tuen: I assumed over-centralization is something you look for in the top of a metagame, not in places where the skill gaps can vary significantly between ranks.

It's not inherently a bad assumption, for the exact reason you've stated.

However, when various iterations of the ban discussion occurred, the topic of dominance at all levels was prevalent. Players believe that utilization of Meta Knight at a local level only required brief knowledge of the character for success. Obviously that's an exaggeration, but it does imply that players do care about dominance issues at those levels as well.

To add to this, we have your statement on varying skill gaps effecting the data. If MK dominated local events even with the risk of combating that variance and running into players who had played much longer than he had, then the result of overpowered-ness would be reinforced.

Wow, that was a long sentence. Hope that made sense!

 

[Flayl]

Could you rephrase that "running into players" bit please? Are you talking about new MK users that overthrow the local champions who used different characters?

 

[Dr. Tuen]

Could you rephrase that "running into players" bit please? Are you talking about new MK users that overthrow the local champions who used different characters?

Yeah, that's the idea. Locals have great variance in skill levels, but if someone from the "mid skill level" overthrew the player considered to be the "local champion" by switching to MK, then it may mean something. It would mean more if that happened frequently at local events across the nation.

If an MK main takes the position of "local champion" and holds it with repeated first place wins, that may also indicate that the natural variation in player skill below him (hopefully in the upward direction) isn't enough to surpass his progress with MK. Again, a result that would mean much more if it is seen frequently.

 

[Sorto]

It's not inherently a bad assumption, for the exact reason you've stated.

However, when various iterations of the ban discussion occurred, the topic of dominance at all levels was prevalent. Players believe that utilization of Meta Knight at a local level only required brief knowledge of the character for success. Obviously that's an exaggeration, but it does imply that players do care about dominance issues at those levels as well.

To add to this, we have your statement on varying skill gaps effecting the data. If MK dominated local events even with the risk of combating that variance and running into players who had played much longer than he had, then the result of overpowered-ness would be reinforced.

Wow, that was a long sentence. Hope that made sense!

I feel its a fair argument to say that only the top level of play should really matter when making a fair ruleset. At the lowest levels of play, spotdodging is hard to beat lol. I know the levels they are considering are higher then that, but overpowered-ness should really matter mainly at the top level. If it is unfair at the top level, then there could be an argument for it being possibly unbeatable. If it is unfair at lower level, the only real argument should be TO GET BETTER.

 

[Dr. Tuen]

I feel its a fair argument to say that only the top level of play should really matter when making a fair ruleset. At the lowest levels of play, spotdodging is hard to beat lol. I know the levels they are considering are higher then that, but overpowered-ness should really matter mainly at the top level. If it is unfair at the top level, then there could be an argument for it being possibly unbeatable. If it is unfair at lower level, the only real argument should be TO GET BETTER.

What about the difference between regionals and nationals? The best example I could think up (that was somewhat realistic) is as follows:

What if the drama at MLG never happened and M2K went on to win the final event? That would count as $12,500 more toward MK. More importantly, because of the interesting events preceding the final event, a Diddy Kong won the event. We can't say for sure if that would be true with M2K back in the mix.

What this ends up meaning is that Diddy earned $12,500 in ONE event, while similar earnings at a national level would require around the vicinity of 10 wins or more. Those 10 events may even have the SAME player base populating the top 20.

So what I'm saying is that at high levels, the scale can effect the results greatly even when the player base may not change significantly.

 

[AlphaZealot]

But you guys never tried increasing the timer (eight minutes doesn't even make any sense) or at least, lowering the stocks and leaving the timer as is.

Less than 1% (actually almost less than one half of 1%) of all matches go to time. The timer would have no effect in terms of decreasing MK dominance.

 

[SaveMeJebus]

Less than 1% (actually almost less than one half of 1%) of all matches go to time. The timer would have no effect in terms of decreasing MK dominance.

That one percent usually happens a high/top level. It's not something you see low level players doing. If you only count high/top level play, the percentage of time outs increases

 

[Dr. Tuen]

That one percent usually happens a high/top level. It's not something you see low level players doing. If you only count high/top level play, the percentage of time outs increases

If I remember right, we only have time data for the MLG events. Those events were pretty much the high/top level players (+anyone else who could afford to go). So I believe that the 1% statement actually applies to high level play alone. We're actually missing low level play data.

 

[SaveMeJebus]

If I remember right, we only have time data for the MLG events. Those events were pretty much the high/top level players (+anyone else who could afford to go). So I believe that the 1% statement actually applies to high level play alone. We're actually missing low level play data.

I'm saying that you usually see it towards the middle and end of the brackets.

 

[Ripple]

That one percent usually happens a high/top level. It's not something you see low level players doing. If you only count high/top level play, the percentage of time outs increases

and do you have anything to back up this claim?

 

[Sorto]

Does anyone even think mk is a game breaking character outside of his planking?

Not over centralized but broken. Don't tell me to define broken, but instead just answer the question and say why he is broken in your opinion.

 

[Dr. Tuen]

I'm saying that you usually see it towards the middle and end of the brackets.

That may be anecdotal experience speaking here. My own anecdotal experience speaks towards the reverse: I've seen more mid-bracket timeouts than I have in grand finals.

I'm unsure if we have a very good grasp on where those match slips fall in each part of their respective tournament brackets, though an analysis could be launched to investigate that. The data is out there.

 

[AlphaZealot]

This functions as a general ranking, but it is one of the weaker methods for relative skill skill rankings out there. Though smash has a problem when trying to move to more accurate methods in that we only have win and loss data. If there were set counts and stock counts for each game, an incredible system could be formulated.

First, in terms of every fighting community in existence, it is the strongest and most complete ranking system. Ever. You can't even find a ranking system for most other communities because their data tracking isn't as evolved as ours. Heck, EVO does not post results past top 8! Let alone the bracket from every pool and full placement all the way down to last place.

Second, stock count would be meaningless. Game count? Might be worth something. But coming close would not increase the accuracy of any ranking system. Chess does not take into account how many pieces are off the board or how 'close' the game is. Again, taking into account games may be OK, taking into account how many stocks were lost? Terrible metric to every use for rankings.

 

[SaveMeJebus]

and do you have anything to back up this claim?

Videos of top level players timing other top level players in the later parts of the bracket.

 

[Dr. Tuen]

First, in terms of every fighting community in existence, it is the strongest and most complete ranking system. Ever. You can't even find a ranking system for most other communities because their data tracking isn't as evolved as ours. Heck, EVO does not post results past top 8! Let alone the bracket from every pool and full placement all the way down to last place.

Second, stock count would be meaningless. Game count? Might be worth something. But coming close would not increase the accuracy of any ranking system. Chess does not take into account how many pieces are off the board or how 'close' the game is. Again, taking into account games may be OK, taking into account how many stocks were lost? Terrible metric to every use for rankings.

Also, to be clear, I'm not trying to undermine the work that has been done on this system. It is widespread, well organized, and concise. Which is awesome.

With respect to the additional details that could be added, it depends on what kind of rankings you are looking at and how they were tabulated. In American Football (at least for college), the point difference in the game actually matters. This would be analogous to adding stock counts to our set data.

For fighting games, our ranking system's power may be true. I'll have to take your word for that. For video games in general, this is far from the truth. LoL, Starcraft, and even World of Warcraft have far superior systems. They, however, have the benefit of having far more data. Many of them hinge on an ELO type system, which in it's truest sense, takes a long time to converge on an appropriate player score. To compensate, they all share excellent online systems which are lag-less enough to allow their results to be indicative of relative player strength. This way, they can collect data every day, while we only get select in-person events to work from. Even then, we don't manage to collect all of that data.

A true improvement to our current system would probably require an overhaul at the website level. If putting the results for a tournament required entering cost numbers, the characters people used and so forth, that data could be mined automatically for further use. Heck, we could even go for the AiB thing and collect TIO brackets, since those are generally the most widely used.

===

ANYWAYS. The original point is that using the rankings we have now for robust analysis should be taken with a grain of salt. While the effort is impressive and good for getting a general idea of who is placed where, I think it introduces enough error to mess up a statistical analysis.

This statement of it's ability to stand up to said analysis is just my opinion though.

 

[SaveMeJebus]

That may be anecdotal experience speaking here. My own anecdotal experience speaks towards the reverse: I've seen more mid-bracket timeouts than I have in grand finals.

I'm unsure if we have a very good grasp on where those match slips fall in each part of their respective tournament brackets, though an analysis could be launched to investigate that. The data is out there.

I haven't really seen or heard of mid level matches go to time. Even then, most of those matches involve two characters that can't approach each other or can't kill each other (a 10 min timer could also help fix this). Pretty much all the time out matches you see in videos are of high/top level players.

 

[DeLux]

@ Tuen

Which BCS computer model uses margin of victory in its system?

I think you're mistaken

 

[Dr. Tuen]

@ Tuen

Which BCS computer model uses margin of victory in its system?

I think you're mistaken

You are correct! A quick google search reveals that margin of victory was only applied in 2/8 BCS computer systems, which intent to phase out the rest. This occurred back in 2002, so it's probably safe to assume that there are none with that compensating factor in them now.

Thanks for the correction. Maybe smash would not benefit from a margin of victory factor... but if it were in my hands and we had that data (for more than one series of events), I'd try it. I like getting hands on with my research :-p.

==

I also seem to have run out of time to chat today. I'll respond to anything directed at me tomorrow! Have a good one guys!

 

[DeLux]

The point of removing margin of victory is that margin as a means of determining more skill is subjective. The only thing that advances you in bracket is winning the set. In theory in a best of 5 set, you could take only 9 stocks while losing 12 stocks and still come out victorious.

So there isn't a direct correlation between who the better player is in a set (aka the winner) in comparison to who took more stocks.

At best, you could make a case for games being a determinant, but definitely not stock differential imo

 

[Tesh]

That whole 1% of matches is bs tho. You guys have data of people that actually let themselves be timed out. Did M2K actually time out Coney or Bizkit? Did Will actually time out Rich Brown?

 

[AlphaZealot]

Tesh: That number is about as non-BS as you could possibly get.

In an analysis (done by Dazwa) of 3,589 games played at MLG last year, only 28 went to time.

Which is roughly 7/10ths of a percent, so LESS than 1% of all games go to time.

Basically: yes, we do, in fact, have the data.

 

[Player-1]

AZ, he's saying that you're not counting matches that definitely would have gone to time if there wasn't a ragequit or some suicidal last ditch attempt to kill them like richbrown vs will. If a game had 1 more second on the clock and the guy was about to time him out so he goes for a latch ditch effort to kill him (some crazy spike) and suicides and the match doesn't go to time, but the threat of being timed out in this is just as real as it going to time.

 

[AlphaZealot]

I see. My point still stands - the number would not be much higher than the number of actual timeouts.

 

[Tesh]

Not really. Do you at least have the number of games that went to 7 minutes? That would be a better indication. Most high profile "timeout" matches don't actually go to time. I can't think of the last time I saw extreme camping actually result in the announcer saying 5-4-3-2-1 TIME.

 

[AlphaZealot]

You realize the entire point of the timer is so that action takes place. Saying "well they only lost their stocks because they had to attack because they were running out of time"...well that is the entire point.

Further, if your point is they only started attacking/trying things because the timer was running out, then it seems probable they would have continued to do nothing if the timer was 10 minutes, and only would begin to take action at the 9 minute mark.

 

[Tesh]

My point still stands the same against yours. You cite actual timeouts when the only difference is that 1 person recklessly goes balls deep in a futile attempt to bring it back. As far as ruleset design there is no difference. The only way broken stalling can prevent the other player from SDing is if you CG them and even then they can ragequit so its not an actual timeout.

The minor difference in exactly how the game ended makes no difference AZ. Your argument fits exactly the same for lowering the timer to like 4 because "well the losing party would just take extreme action at 3 minutes and it would affect nothing".

 

[AlphaZealot]

To be honest I'm not particularly sure what we are arguing about.

 

[John12346]

Well, I should put something out there, then, since we're all sitting around. I found a flaw with some data I had put forth earlier just now.

I'm not getting into this argument, but Sorto, only 11.69% of Brawl players use MK in tournament play. A mere 7.94% actually main him.

Someone had requested some data from me regarding Rajam's charts, so when I rechecked the numbers, I noticed I made a massive error. When I did the initial calculations, I assumed dividing by the total number of players on Rajam's charts(4609) would do the trick. But then I found out about 1500 players(randoms) did not have their characters accounted for, giving us a total playerbase of 2760 instead.

Recalculating, the correct values are:
Approximate percentage of people who main MK: 366 out of 2760 - 13.26%
Approximate percentage of people who use MK: 539 out of 2760 - 19.53%

Make of it what you will, but here you go.

 

[theunabletable]

Less than 1% (actually almost less than one half of 1%) of all matches go to time. The timer would have no effect in terms of decreasing MK dominance.

this is a bit narrow to say, given the results we have.

Sure .5% of matches at MLG went to time, but I know from personal experience, I get very few timeouts on stages like RC or Brinstar, I usually just win the match normally. However I only win normally because the opponent tries risky things to try and get back the lead because the timer is short.

It's kinda shortsighted to imply "less than 1% of matches go to time, therefore less than 1% of matches are even affected by the timer".

It's not because the timer made the game more fighting based or anything like that, it's not fulfilling any goal like that. It's that the risk:reward in that situation for my opponent becomes one where he either automatically loses, and it takes a really long time, and he probably gets more mentally drained, or he almost definitely loses earlier, and might be able to land a few hits.

Recalculating, the correct values are:
Approximate percentage of people who main MK: 366 out of 2760 - 13.26%
Approximate percentage of people who use MK: 539 out of 2760 - 19.53%

Make of it what you will, but here you go.

wait that's IT?

That's a LOT less than I thought lol.

 

[ElDominio]

this is a bit narrow to say, given the results we have.

Sure .5% of matches at MLG went to time, but I know from personal experience, I get very few timeouts on stages like RC or Brinstar, I usually just win the match normally. However I only win normally because the opponent tries risky things to try and get back the lead because the timer is short.

It's kinda shortsighted to imply "less than 1% of matches go to time, therefore less than 1% of matches are even affected by the timer".

It's not because the timer made the game more fighting based or anything like that, it's not fulfilling any goal like that. It's that the risk:reward in that situation for my opponent becomes one where he either automatically loses, and it takes a really long time, and he probably gets more mentally drained, or he almost definitely loses earlier, and might be able to land a few hits.

What???
As in,

???

I only win normally because the opponent tries risky things to try and get back the lead

Hence the reason it didn't go to time out...

because the timer is short.

What?????????

If the timer were longer, then
1) He wouldn't do those risky things, the game would last a few minutes more, and a time limit
2) He dies, only a few minutes later, due to him doing the exact same thing only later
or
3) you kill him at 8:44

It's still 44 seconds longer than normal -_-

 

[Sorto]

What constitutes over-centralization? Can two characters over-centralize a meta-game?

Current argument by URC:
MK is over centralized because he takes in 45.91% of the money.

My Discussion:
Diddy and Snake take in 10.06% and 16.57% of the tournament money respectively.
This means diddy and snake together take in 26.63% of total tournament winnings.
This is 49.23% of the all the remaining money not taken in by Meta Knight. Isn't it possible that these two characters are over-centralize in comparison to the remaining cast using the current argument described above.

Just food for thought...



Original Data From John12346:
Meta Knight
Money won from 2011 tournaments(derives from Average category on my charts): $42047.94 out of $91588.44 - 45.91%

Diddy
Money won from 2011 tournaments(derives from Average category on my charts): $9217.71 out of $91588.44 - 10.06%

Snake
Money won from 2011 tournaments(derives from Average category on my charts): $15173.03 out of $91588.44 - 16.57%

Simple Mathematical Equation:
100*((10.06+16.57)/(100-45.91))=49.23



**I know money earned by each character may not be mutually exclusive. But that data could be figured out and I believe that it is likely that the payout amounts would still reflect the idea I described above.

 

[SaveMeJebus]

If you look at the data for MK banned tournaments, Snake's percentage is already at about 20%. This could go higher as soon as TOs start hosting MK banned Nationals

 

[John12346]

Bleh, response from a different thread, w/e

Anyway, 20% is not really a lot considering that, during the fourth MK ban debate all those years ago, anti-ban dismissed 30% as not enough dominance...

Also, on a cumulatively collected scale, MK had his highest percentage(for "Average") at 51.0198% right at the second week of 2011 - 1/8/11. As you can see, after all this time, it has decreased to 45.9097%. The cumulative results for dominant characters will generally always decrease from their initial amounts given from a low amount of tournies, and level off to their true values as you collect more and more data.

It happened to MK(although MK retained very large numbers regardless); I'm sure it's going to end up happening to Snake as well.

What Sorto said is legit enough, but it's only assuming that ALL of MK's money goes to Diddy and Snake, and I'm 99% sure that other characters are going to end up receiving sizable portions of MK's winnings in an MK banned metagame too...

 

[Smooth Criminal]

What Sorto said is legit enough, but it's only assuming that ALL of MK's money goes to Diddy and Snake, andI'm 99% sure that other characters are going to end up receiving sizable portions of MK's winnings in an MK banned metagame...

You don't know that, John. Not for sure anyway.

This is why overcentralization arguments are bad. There was much better criterion for you guys to base this ban off of. Instead, ya'll throw it on to the back burner and laud your numbers rather than focus on the fact that MK "breaks" your game and/or your rulesets.

I wonder, then, what it's going to be like when another character starts bringing in the bulk of the money...? Are we going to take a hard look at him or her and opt 'em out for another community ruling, based on overcentralization...?

Smooth Criminal

 

[Sorto]

Bleh, response from a different thread, w/e


What Sorto said is legit enough, but it's only assuming that ALL of MK's money goes to Diddy and Snake, and I'm 99% sure that other characters are going to end up receiving sizable portions of MK's winnings in an MK banned metagame too...

This is true. But I am saying that in the current metagame and ruleset that the following is true:
1. If it is true that MK over centralizes the cast.
2. Then it is true that diddy and snake overcentralize the remainder of the cast outside of MK.

It is a general pattern in any competitive fighter that higher tiered characters will win more tournies and earn more money then lowered tiered ones. The higher tiers will always earn a large percent of the money pot. I did this to show the flaw with creating a percentage and claiming it as a sign of OVERCENTRALIZATION. Then using this somewhat arbitrary value/belief as a reason for banning.

As I said in the first post,

This is just food for thought...

 

[John12346]

Well, yeah, money won isn't really an indicator of overcentralization, amount of usage is, of course. Other pro-banners seem to say that ~20% usage, as evidenced by Rajam and Ripple's lists, are. I'm not 100% sure how I feel on this point, but that's what they say.

But yeah, what you said is some good reasoning. When we just remove MK's money flat out, Snake and Diddy's cash combined rises to half of all tournament money, for the most part. It's definitely something to consider.