|
Hong Kong9179 Posts
On September 20 2012 22:22 ArcticFox wrote:
Swiss style would, again, mean that there's no games worth watching until either late Friday night, or early Saturday morning, when the 4-0 and 5-0 people start facing off against each other. And once again, when the top 32/64/128 were seeded into a bracket, there would be a couple of rounds before your big names started playing against each other again. It would almost surely end up being the best player who emerged victorious, but there would be a lot of "dead air" where nobody is watching inbetween..
This might only be a problem with the first two randomly assigned and preset rounds. Afterwards, high-high power matching begins to ensure that the top people consistently hit each other in a race to the top. Nearer the end, we get to see the break rounds, where people on the cusp of making it into elimination contention can be featured.
And only the top players make it in. If they are not big names so be it. Our community has to examine whether it is strength of play which matters or big names. Modified-Swiss procedure at least lets any player who breaks out prove their worth, unlike other tournament procedures where a lot of luck and intervention can get you a long way.
|
On September 20 2012 22:49 itsjustatank wrote:Show nested quote +On September 20 2012 22:22 ArcticFox wrote:
Swiss style would, again, mean that there's no games worth watching until either late Friday night, or early Saturday morning, when the 4-0 and 5-0 people start facing off against each other. And once again, when the top 32/64/128 were seeded into a bracket, there would be a couple of rounds before your big names started playing against each other again. It would almost surely end up being the best player who emerged victorious, but there would be a lot of "dead air" where nobody is watching inbetween.. This might only be a problem with the first two randomly assigned and preset rounds. Afterwards, high-high power matching begins to ensure that the top people consistently hit each other in a race to the top. Nearer the end, we get to see the break rounds, where people on the cusp of making it into elimination contention can be featured. And only the top players make it in. If they are not big names so be it. Our community has to examine whether it is strength of play which matters or big names. Modified-Swiss procedure at least lets any player who breaks out prove their worth, unlike other tournament procedures where a lot of luck and intervention can get you a long way.
Fair or not, the big names are what gets people to tune in and watch.
Also, I'm not sure if you've seen the inner workings of the MLG Open Bracket, which is what a full Modified-Swiss system would be, but the first round of the Open Bracket with 256 players in it takes 4 hours to get through, by rotating people in and out. Which means Round 2 of a Swiss system would *also* take another 4 hours. That means it would take the entirety of Day 1 to get through 2 rounds, meaning it would be Saturday morning before we saw our first 2-0 vs. 2-0 matches come up. Either that, or it would require MLG to double the amount of machines they have, which would double their cost (or require them to convince Alienware to pony up for twice as many PCs), and require them to find twice as much floorspace for just SC2.
Again, for fairness and making sure the best players rise to the top, I like this idea. For practicality and viewability? It would be really hard to sell.
|
Canada2480 Posts
O do understand the points being made in favor of double elimination tournaments for broadcasters etc. From a competitive standpoint, I feel like the single elimination format works the best. If you win you win if you lose you lose. What's unfair about that?
|
Hong Kong9179 Posts
On September 21 2012 00:55 ArcticFox wrote:Show nested quote +On September 20 2012 22:49 itsjustatank wrote:On September 20 2012 22:22 ArcticFox wrote:
Swiss style would, again, mean that there's no games worth watching until either late Friday night, or early Saturday morning, when the 4-0 and 5-0 people start facing off against each other. And once again, when the top 32/64/128 were seeded into a bracket, there would be a couple of rounds before your big names started playing against each other again. It would almost surely end up being the best player who emerged victorious, but there would be a lot of "dead air" where nobody is watching inbetween.. This might only be a problem with the first two randomly assigned and preset rounds. Afterwards, high-high power matching begins to ensure that the top people consistently hit each other in a race to the top. Nearer the end, we get to see the break rounds, where people on the cusp of making it into elimination contention can be featured. And only the top players make it in. If they are not big names so be it. Our community has to examine whether it is strength of play which matters or big names. Modified-Swiss procedure at least lets any player who breaks out prove their worth, unlike other tournament procedures where a lot of luck and intervention can get you a long way. Fair or not, the big names are what gets people to tune in and watch. Also, I'm not sure if you've seen the inner workings of the MLG Open Bracket, which is what a full Modified-Swiss system would be, but the first round of the Open Bracket with 256 players in it takes 4 hours to get through, by rotating people in and out. Which means Round 2 of a Swiss system would *also* take another 4 hours. That means it would take the entirety of Day 1 to get through 2 rounds, meaning it would be Saturday morning before we saw our first 2-0 vs. 2-0 matches come up. Either that, or it would require MLG to double the amount of machines they have, which would double their cost (or require them to convince Alienware to pony up for twice as many PCs), and require them to find twice as much floorspace for just SC2. Again, for fairness and making sure the best players rise to the top, I like this idea. For practicality and viewability? It would be really hard to sell.
Infrastructure inefficiencies that limit even the conduction of a flawed system are the tournament organizer's problem in your example, not mine. You assume a steady state that is unchangeable. If a tournament organizer needs to have extra computers or admins to get a proper tournament procedure in place, they should do it because it isn't impossible to do especially at the scale of event you are talking about.
|
I'm sick of tournament organizers making lopsided groups. The MLG bracket was 3 ridiculously hard brackets and 1 really easy one.
Now in Dreamhack Grubby wins his group 3-0 and his reward for doing good in the first group stage is to be paired up with Stephano and Taeja in the 2nd group stage. That doesn't make any sense at all.
|
I realise I'm obviously late to the party, but I just happened to stumble upon this blog today and I somehow find the mechanics of tournament formats intriguing, so I'll add my two cents nonetheless. More to the point, two observations:
First, with Starcraft tournaments being the way they are, with always just two players who always either win or lose, wouldn't the format in the OP basically turn into a double, triple or N-multiple elimination bracket, depending on the point of cut-off, with the only exception being that the seed from the winners' bracket to the losers' bracket are not mixed into the bracket itself (as usual), but rather played high-high (also depicted in the second image)? I've made an illustration to show it. (I'm crap at using photoshop and stuff, so I did it the old-fashioned way, sorry).
![[image loading]](http://i.imgur.com/de7AY.jpg)
Second, since win and lose are the only possible outcomes of a match, there's actually a 100% predictability to the player's path up until the point where there's an odd number of players with an equal score, which shouldn't happen for a very long time, given a huge pool of initial players, and probably shouldn't happen at all before the Tabulation format is left in favour of the single-elimination playoffs. Another illustration (which also shows the point where predictability fails, in round 4 given 16 initial players and a cut-off at an equivalent of double elimination). A box equals matches and not players:
![[image loading]](http://i.imgur.com/xk5O5.jpg)
I don't really mean to argue either for or against the Tabulation format, I just wanted to point out how it's not that different from N-multiple elimination and also that it's not that unpredictable and can easily (if you're well-prepared, at least) be done without spreadsheets or computers. Also, with it being predictable, players can prepare their games better, although one completely loses the excitement of having the next competitors announced.
All this given that there's no finer point of it that I missed and that I analyzed this correctly, of course 
Edit: Also, as long as there's 100% predictability, there's no need to wait for everyone to finish their round before moving onto the next.
|
Hong Kong9179 Posts
Your analysis ignores the crucial factor of how one gets seeded into a single- double- or n- elimination bracket matters quite heavily on how you fare in the tournament. If you already have proper seeds going into a tournament, the tournament should of course be a single-elimination tournament based on your seeds. However, if you do not have seeds going into the tournament, random assignment to an elimination bracket is unacceptable. Group stage into elimination (the current regime) and tabulation seek to remedy this problem.
On September 21 2012 01:10 swanized wrote:
O do understand the points being made in favor of double elimination tournaments for broadcasters etc. From a competitive standpoint, I feel like the single elimination format works the best. If you win you win if you lose you lose. What's unfair about that?
The above also answers this question as well. If you read the text however, I do not advocate double-elimination at all.
|
I've been thinking about a good format for my sports club yearly championship (table tennis). So far, we used round-robins with knockout phases after (but a few separate tourneys (A,B,C) for different levels as skill differences are huge). So I noted a couple things I wanted to improve:
- anyone should be able to win the whole thing
- the number of rounds is limited as everything is to be played on 1 day/night
- the number of participants is unknown in advance (until the day before)
- every participant should play a 'good' number of matches, nobody wants to play 2 games and then be done
- there should be a deciding final
This thread actually gave me an idea, why don't we play a Swiss system with all the players (say 5 rounds for up to 32 players), then release the players in stages into a single-elim tournament (bottom 8 vs 8, then next 4, then another 4, etc) while the rest continue with the Swiss? You can theoretically still win after the swiss part, you always play at least 6 games, and you play mostly opponents that are near your level.
I'm most worried about the mechanics of the Swiss style tournament on the spot (having computers there is awkward) and the middle section of the Swiss result table not being very accurate. What dyou guys think about this?
|
Hong Kong9179 Posts
On November 06 2012 19:23 aseq wrote:
I've been thinking about a good format for my sports club yearly championship (table tennis). So far, we used round-robins with knockout phases after (but a few separate tourneys (A,B,C) for different levels as skill differences are huge). So I noted a couple things I wanted to improve:
- anyone should be able to win the whole thing
- the number of rounds is limited as everything is to be played on 1 day/night
- the number of participants is unknown in advance (until the day before)
- every participant should play a 'good' number of matches, nobody wants to play 2 games and then be done
- there should be a deciding final
This thread actually gave me an idea, why don't we play a Swiss system with all the players (say 5 rounds for up to 32 players), then release the players in stages into a single-elim tournament (bottom 8 vs 8, then next 4, then another 4, etc) while the rest continue with the Swiss? You can theoretically still win after the swiss part, you always play at least 6 games, and you play mostly opponents that are near your level.
I'm most worried about the mechanics of the Swiss style tournament on the spot (having computers there is awkward) and the middle section of the Swiss result table not being very accurate. What dyou guys think about this?
Not sure why you would continue with the Swiss, except to let eliminated players have something to do after being effectively eliminated. If you are going to continue with the Swiss, then it should just be pure Swiss because that is the absolute best and most fair way to determine a winner. If you are going to use Swiss for seeding, then what I described in the OP works.
|
Yup, my reasons to do it like that would be:
- The higher-ranked players need something to do while the rest play their knockout rounds. Can't have them waiting for up to 2 hours.
- I do want the knockout rounds, because I want to have a grand final at the end. Swiss systems just don't have good endings - things can even be decided before the last game. We're not world class level, so finding the best player is important, but just as important is an exciting tournament finale.
I also looked into McMahon, which seems interesting as we don't have that much time (but might raise some complaints from people). I'm not certain what you pick yet, I have plenty of time left, only I'd hoped this thread would have picked up a lot better to find more inspiration  .
|
Hong Kong9179 Posts
I don't think you understand the system. There are two parts of the system I describe here, for clarity I'm going to approach it as a two-day tournament: 1) preliminaries using a swiss system to determine seeding for 2) the next day's elimination-brackets. If you didn't make it to the elim rounds on day 2, there is nothing for you to do because you are out of the tournament.
|
On November 09 2012 14:38 itsjustatank wrote:
I don't think you understand the system. There are two parts of the system I describe here, for clarity I'm going to approach it as a two-day tournament: 1) preliminaries using a swiss system to determine seeding for 2) the next day's elimination-brackets. If you didn't make it to the elim rounds on day 2, there is nothing for you to do because you are out of the tournament.
I perfectly understand that. And that's precisely why I'm looking at making some changes. As Swiss doesn't yield too great results in the middle of the pack, #9 after the Swiss (in case of 3rounds of knockout) is going to feel left out. Also, I don't want more than half of my players being done and out after half the tournament. That's why I'm proposing to extend the knockout to all the players (like the MLG, but not as long). And because of the gap in time, then get the top players to continue the Swiss.
|
Hong Kong9179 Posts
On November 09 2012 23:35 aseq wrote:Show nested quote +On November 09 2012 14:38 itsjustatank wrote:
I don't think you understand the system. There are two parts of the system I describe here, for clarity I'm going to approach it as a two-day tournament: 1) preliminaries using a swiss system to determine seeding for 2) the next day's elimination-brackets. If you didn't make it to the elim rounds on day 2, there is nothing for you to do because you are out of the tournament. I perfectly understand that. And that's precisely why I'm looking at making some changes. As Swiss doesn't yield too great results in the middle of the pack, #9 after the Swiss (in case of 3rounds of knockout) is going to feel left out. Also, I don't want more than half of my players being done and out after half the tournament. That's why I'm proposing to extend the knockout to all the players (like the MLG, but not as long). And because of the gap in time, then get the top players to continue the Swiss.
You cannot fairly seed an elimination tournament in this way. People who are in the bottom and get sent to the elim bracket have less accurate placements in the tournament that they could have gained back through more rounds in swiss. Players at the top are punished by having to play fundamentally more games than people being sent to the elimination bracket early. All of this in the name of 'letting people have things to do.' It's not a fair or efficient system, it is needlessly complicated for an advantage that hasn't been articulated.
|
On November 09 2012 23:43 itsjustatank wrote:Show nested quote +On November 09 2012 23:35 aseq wrote:On November 09 2012 14:38 itsjustatank wrote:
I don't think you understand the system. There are two parts of the system I describe here, for clarity I'm going to approach it as a two-day tournament: 1) preliminaries using a swiss system to determine seeding for 2) the next day's elimination-brackets. If you didn't make it to the elim rounds on day 2, there is nothing for you to do because you are out of the tournament. I perfectly understand that. And that's precisely why I'm looking at making some changes. As Swiss doesn't yield too great results in the middle of the pack, #9 after the Swiss (in case of 3rounds of knockout) is going to feel left out. Also, I don't want more than half of my players being done and out after half the tournament. That's why I'm proposing to extend the knockout to all the players (like the MLG, but not as long). And because of the gap in time, then get the top players to continue the Swiss. You cannot fairly seed an elimination tournament in this way. People who are in the bottom and get sent to the elim bracket have less accurate placements in the tournament that they could have gained back through more rounds in swiss. Players at the top are punished by having to play fundamentally more games than people being sent to the elimination bracket early. All of this in the name of 'letting people have things to do.' It's not a fair or efficient system, it is needlessly complicated for an advantage that hasn't been articulated.
Okay. I did mean that people who stay in the Swiss play 1 round per round the people in the elimination bracket play. So they don't play more games (you play the same amount until you get kicked out in elims). Like in the MLG, people who end lower in the bracket have a long(er) way to go in the elimination part. True, it may not have any benefits finding the eventual winner, but I don't see how this would be 'less fair' than just having your Swiss into top 8 elim method.
|
Hong Kong9179 Posts
On November 10 2012 01:22 aseq wrote:Show nested quote +On November 09 2012 23:43 itsjustatank wrote:On November 09 2012 23:35 aseq wrote:On November 09 2012 14:38 itsjustatank wrote:
I don't think you understand the system. There are two parts of the system I describe here, for clarity I'm going to approach it as a two-day tournament: 1) preliminaries using a swiss system to determine seeding for 2) the next day's elimination-brackets. If you didn't make it to the elim rounds on day 2, there is nothing for you to do because you are out of the tournament. I perfectly understand that. And that's precisely why I'm looking at making some changes. As Swiss doesn't yield too great results in the middle of the pack, #9 after the Swiss (in case of 3rounds of knockout) is going to feel left out. Also, I don't want more than half of my players being done and out after half the tournament. That's why I'm proposing to extend the knockout to all the players (like the MLG, but not as long). And because of the gap in time, then get the top players to continue the Swiss. You cannot fairly seed an elimination tournament in this way. People who are in the bottom and get sent to the elim bracket have less accurate placements in the tournament that they could have gained back through more rounds in swiss. Players at the top are punished by having to play fundamentally more games than people being sent to the elimination bracket early. All of this in the name of 'letting people have things to do.' It's not a fair or efficient system, it is needlessly complicated for an advantage that hasn't been articulated. Okay. I did mean that people who stay in the Swiss play 1 round per round the people in the elimination bracket play. So they don't play more games (you play the same amount until you get kicked out in elims). Like in the MLG, people who end lower in the bracket have a long(er) way to go in the elimination part. True, it may not have any benefits finding the eventual winner, but I don't see how this would be 'less fair' than just having your Swiss into top 8 elim method.
It's less fair because the system makes no sense in terms of determining the real winner of a tournament. You have people playing meaningless games and tiring themselves out in extra Swiss rounds just so 'they have something to do.'
In particular, you lose the very reason why Swiss rounds are conducted in preliminaries in the first place. Determining accurate and fair seeding prior to an elimination bracket would provide substantial benefits for the top seeded players who proved those seeds in preliminary play. You eliminate that.
|
Well, I would like to know if you have any better alternatives for my tournament, then. So far you're only saying it makes no sense, but providing little reasoning.
In Swiss, ranking gets better with the amount of games you play. Since you only play people at approx your level, even cutting of half of the players and continuing will IMPROVE the ranking of the remainging players, as higher players don't have anything to do with lower players anyway. The longer you continue, the closer you're getting to a round-robin, which is ideal, but not feasible timewise. So getting them to play on in the Swiss system will get better results.
I read this part about 8 times, but I don't get what you're trying to say in your second paragraph. How am I stopping top players from getting the best seeds? I'm not. The people who stay in the Swiss for longest get the best seeds, and also have to play the fewest elimination games. That's a substantial benefit.
|
Hong Kong9179 Posts
You don't play people approximate to your level in your system because you arbitrarily send a group of players to the elimination bracket in the middle of Swiss play. This fundamentally changes who can hit who in further Swiss rounds and the dynamics of the games that follow. My alternative is the system I present in this blog time and time again, which is to conduct a fixed set of Swiss rounds and seed a single-elimination tournament based on performance in those rounds. I don't know why you are dead-set on adding entropy into the Swiss system just so people 'have something to do.' It tanks the benefits of using Swiss in the preliminaries in the first place. At the point where any tournament would consider using your system, it would be better off just randomly seeding a double-elimination bracket.
|
Okay, your system is more fair, i'll give you that. Mind you, is isn't a big change i'm proposing. I think you're mistaken saying it's about as good as double-elim, since it's simply not that different from yours. Discrete advantages would be:
- There would be no #1 vs #64 games in the first round of elim...which are boring. My proposal gets lower skilled players to beat up on the next guy each time.
- I'm not really breaking into the Swiss system, removing players from it. I'm adding more rounds after it for the top players (yes, which are optional).
- Top players can afford a slip-up at anytime but the last 2 or 3 games, instead of 5.
So, removing the extra swiss rounds (which seem to madden you greatly ), but making the single-elim a bit skewed (adding players in multiple stages in some yet-to-be-decided manner), what would you say to that?
|
Hong Kong9179 Posts
On November 11 2012 12:21 aseq wrote:Okay, your system is more fair, i'll give you that. Mind you, is isn't a big change i'm proposing. I think you're mistaken saying it's about as good as double-elim, since it's simply not that different from yours. ... - I'm not really breaking into the Swiss system, removing players from it. I'm adding more rounds after it for the top players (yes, which are optional). ... So, removing the extra swiss rounds (which seem to madden you greatly ), but making the single-elim a bit skewed (adding players in multiple stages in some yet-to-be-decided manner), what would you say to that?
I still don't understand their function for the elimination tournament to follow. Right now it just sounds like extra showmatches to be played before elimination matches start. You yourself don't seem to particularly be able to explain how to add people to the elimination bracket in stages.
The reason why what you propose is not optimal is that someone can end up going 0-2 in the tournament's first two rounds, but fight back to 4-2 and break (assuming six rounds total). If you send them to the elimination bracket early based on that initial bad performance, you rob them of the chance to redeem themselves in later Swiss rounds. And this is made uniquely worse, because the first set of rounds for Swiss are randomly assigned. Only through a full set of Swiss rounds can we accurately determine seeding into an elimination bracket fairly.
You also lose the storyline generated when you show break rounds in the later stages of the tournament, watching people fight for a spot in the elimination bracket.
On November 11 2012 12:21 aseq wrote:
- There would be no #1 vs #64 games in the first round of elim...which are boring. My proposal gets lower skilled players to beat up on the next guy each time.
One of the departures my system takes from standard Swiss is that it does not clear the entire pool to the elimination bracket (although I guess clearing the entire field can be done using my system, if so desired). In the activity from which I am porting this procedure over, we generally clear about half of the player pool, eliminating every one else below that point from tournament contention.
In this way, top seed will not hit the absolute worst person in the tournament, but rather they will hit the lowest person to clear (who comes from the middle of the pack).
On November 11 2012 12:21 aseq wrote:
- Top players can afford a slip-up at anytime but the last 2 or 3 games, instead of 5.
Top players in the last rounds of Swiss don't really slip up. They will clear no matter what in my system. They may lose top seed, but the whole point of power matching is to determine whether players deserve to be in elimination bracket contention, and to determine where they deserve to be seeded.
|
Okay, clear. I think I want everyone to take part in the elim too, so that's why this may not be the best idea, then. I also think that for my purposes, only a seed is too small a reward for ending on top after Swiss (esp. so when the field is very level). In many tournaments, the Swiss by itself determines the winner already. That's why I was thinking of the layered approach.
Thanks anyway for the write-up and explanation, I'll think about it some more and discuss with others.
|
|
|
|
|
|
EPT
