|
So I'm normally a super huge lurker, but after reading through a lot of TL posts about the same topic, I'm finally making a thread about something that seriously grinds my gears.
EXTENDED SERIES IS NOT DOUBLE ELIMINATION.
Holy crap I don't know why people are so confused between the two concepts. Any time there's a double elimination tournament people on Reddit/TL/WP/stream will say something that amounts to: "WTF IS THIS EXTENDED SERIES BS" when in the finals the WB winner needs to win 1 BoX and the LB winner needs to win 2 BoX like, say, EU Blizz invitational. The current extended series rules are bad enough to hate on its own merits, there's no need to drag in double elim, which is a totally fine tournament format.
(I'm going to go into a huge side-rant about extended series here, ignore if you want to focus on the main rant of the difference between extended series and double elim)
Extended series is when in a double elimination tournament, two players that have previously met have a score that carries over from their previous match. Ie. A plays B and wins 2-1. They meet again. Match starts off 2-1 as a Bo7. A still only needs to win 2 games, while B needs to win 3, as opposed to straight double-elim, where the second match is just a Bo3. The only time extended series enters into the Grand Finals is if A and B both get there. Then they play an extended series as if they had met in the losers, with the winner of that extended series being the winner of the tournament. That's the problem with extended series as MLG has it. Assuming extended series is fair (debatable), if A and B meet in losers, that's fine, because they've both already lost a series so the loser of that extended series deserves to be eliminated. But if A and B meet in Grand Finals, A being up 2 games with B needing 3-4 is the same (or worse) advantage as A would have received from being WB winner anyway. I understand that MLG is saying A is the worse player compared to B if A loses, but in that case, A has received no advantage from being the WB winner, and there is still a missing Bo3. In order for the format to be fair and logically consistent, there needs to be an extended series of the extended series if A loses. That means it's a Bo11 (first to 6), with A and B up however many they won before. This is logically consistent, ie. B could go 0-2, then 4-3 in the extended series; B would therefore need 2 more maps to win in the extended extended series, while A needs 3 more because he's down a map from the first extended series. The same logic that we accepted as an assumption (that we want to judge the best player between them in one full series) used to justify normal extended series can be used to justify the extended extended series, and the WB winner (A) receives his advantage of having a safety net of 1 loseable match. That would mean LB winner would need to win 3-4 games before WB winner wins 2, then 2 games before WB winner wins 3-4. LB winner would therefore need to win 5-6 games total, while WB winner needs 2-4. "True" extended series would make matches longer, not shorter. Whether that's a good thing or not is debateable, as is the logic that suggests its use. Current MLG rules are unequivocally unfair for WB winners, however. In the case of extended series in grand finals (thankfully rare), WB winners receive no advantage or are penalized for coming from winners.
However, needing LB winner to win 2 Bo3 isn't extended series by any stretch. That's called double elimination. A previous series isn't being extended when, in the Grand Finals, the player who hasn't lost yet needs to lose twice to someone who has. They're playing two separate series. The Grand Finals with MLG-style extended series only occurs if the WB winner knocked the LB winner to losers in the first place. How did the concept of double elimination and extended series get confused and mixed so thoroughly in so many people's heads? It makes reading posts about "extended series" super confusing especially because everyone assumes the others are using it the same way they are, only some of them aren't. For example, this post. Is he talking about double elim finals or extended series finals? Who knows? Well, presumably he does, and I can guess that he's talking about double elim because MLG-style extended series is actually a boon to the loser bracket winner so it shouldn't be more anticlimatic, but the fact is, this shouldn't be so confusing. It only gets worse when people quote him and continue the debate. At this point, I can't tell if they're confused about what they're arguing for, or just wrong.
Should tournaments use extended series? Who knows, but hopefully not like MLG does it. Is the person coming from LB needing to win 2 BoX fair (it is) or good for viewer excitement (it is arguably not)? That is a whole different question. There's a recent thread about it! Check it out! But please, extended series is not double elim. Next time there's a grand finals in a double elim tournament, whine about double elims, not about extended series (unless it was actually used, and changed something). And when you're arguing about either of the terms, use them correctly! Thanks in advance.
|
By definition, extended series is a double elimination. However, double elimination is not necessarily an extended series.
Anyways, I don't understand why many people like the double elimination format. Granted, a double-elim bo3 is superior to single-elim bo3, but I much prefer single-elim bo5. In my opinion, if you lose a bo5 to a "weaker" player, you don't deserve to be in the tournament anymore.
Anyways, extended series is confusing, but not really a big deal. I think what's more important for the MLG format is:
- The points distribution system (i.e. the people seeded will take ages to fall out).
- The "marathon-like" nature of the event which is too taxing.
|
On August 23 2011 12:01 VikingKong wrote:
(I'm going to go into a huge side-rant about extended series here, ignore if you want to focus on the main rant of the difference between extended series and double elim)
Extended series is when in a double elimination tournament, two players that have previously met have a score that carries over from their previous match. Ie. A plays B and wins 2-1. They meet again. Match starts off 2-1 as a Bo7. A still only needs to win 2 games, while B needs to win 3, as opposed to straight double-elim, where the second match is just a Bo3. The only time extended series enters into the Grand Finals is if A and B both get there. Then they play an extended series as if they had met in the losers, with the winner of that extended series being the winner of the tournament. That's the problem with extended series as MLG has it. Assuming extended series is fair (debatable), if A and B meet in losers, that's fine, because they've both already lost a series so the loser of that extended series deserves to be eliminated. But if A and B meet in Grand Finals, A being up 2 games with B needing 3-4 is the same (or worse) advantage as A would have received from being WB winner anyway. I understand that MLG is saying A is the worse player compared to B if A loses, but in that case, A has received no advantage from being the WB winner, and there is still a missing Bo3. In order for the format to be fair and logically consistent, there needs to be an extended series of the extended series if A loses. That means it's a Bo11 (first to 6), with A and B up however many they won before. This is logically consistent, ie. B could go 0-2, then 4-3 in the extended series; B would therefore need 2 more maps to win in the extended extended series, while A needs 3 more because he's down a map from the first extended series. The same logic that we accepted as an assumption (that we want to judge the best player between them in one full series) used to justify normal extended series can be used to justify the extended extended series, and the WB winner (A) receives his advantage of having a safety net of 1 loseable match. That would mean LB winner would need to win 3-4 games before WB winner wins 2, then 2 games before WB winner wins 3-4. LB winner would therefore need to win 5-6 games total, while WB winner needs 2-4. "True" extended series would make matches longer, not shorter. Whether that's a good thing or not is debateable, as is the logic that suggests its use. Current MLG rules are unequivocally unfair for WB winners, however. In the case of extended series in grand finals (thankfully rare), WB winners receive no advantage or are penalized for coming from winners.
This is so unclear. I honestly have no idea what you're trying to say, even after reading it multiple times.
|
On August 23 2011 12:17 Azzur wrote:
By definition, extended series is a double elimination. However, double elimination is not necessarily an extended series.
Anyways, I don't understand why many people like the double elimination format. Granted, a double-elim bo3 is superior to single-elim bo3, but I much prefer single-elim bo5. In my opinion, if you lose a bo5 to a "weaker" player, you don't deserve to be in the tournament anymore.
Anyways, extended series is confusing, but not really a big deal. I think what's more important for the MLG format is:
- The points distribution system (i.e. the people seeded will take ages to fall out).
- The "marathon-like" nature of the event which is too taxing.
Yeah, that's exactly my point. People equate the two, and it makes their arguments confusing. People whine about extended series when there were no series being extended, or when the context is ambiguous. Extended series MLG style is no big deal until the grand finals, like I said in my huge rant in the middle. At that point, the winner's bracket finalist gets no advantage for coming from winner's if it's an extended series.
As for double elim/single elim, I think double is good because it mitigates first round crazy matchups, like ryung/mma in the round of 64 in IPL qual #1. I like Chill's idea that it switches to a single elim starting Ro16. Avoids the bad-luck seeding, and increases viewer excitement since every match will eliminate someone.
On August 23 2011 12:23 vVvTime wrote:Show nested quote +On August 23 2011 12:01 VikingKong wrote:
(I'm going to go into a huge side-rant about extended series here, ignore if you want to focus on the main rant of the difference between extended series and double elim)
Extended series is when in a double elimination tournament, two players that have previously met have a score that carries over from their previous match. Ie. A plays B and wins 2-1. They meet again. Match starts off 2-1 as a Bo7. A still only needs to win 2 games, while B needs to win 3, as opposed to straight double-elim, where the second match is just a Bo3. The only time extended series enters into the Grand Finals is if A and B both get there. Then they play an extended series as if they had met in the losers, with the winner of that extended series being the winner of the tournament. That's the problem with extended series as MLG has it. Assuming extended series is fair (debatable), if A and B meet in losers, that's fine, because they've both already lost a series so the loser of that extended series deserves to be eliminated. But if A and B meet in Grand Finals, A being up 2 games with B needing 3-4 is the same (or worse) advantage as A would have received from being WB winner anyway. I understand that MLG is saying A is the worse player compared to B if A loses, but in that case, A has received no advantage from being the WB winner, and there is still a missing Bo3. In order for the format to be fair and logically consistent, there needs to be an extended series of the extended series if A loses. That means it's a Bo11 (first to 6), with A and B up however many they won before. This is logically consistent, ie. B could go 0-2, then 4-3 in the extended series; B would therefore need 2 more maps to win in the extended extended series, while A needs 3 more because he's down a map from the first extended series. The same logic that we accepted as an assumption (that we want to judge the best player between them in one full series) used to justify normal extended series can be used to justify the extended extended series, and the WB winner (A) receives his advantage of having a safety net of 1 loseable match. That would mean LB winner would need to win 3-4 games before WB winner wins 2, then 2 games before WB winner wins 3-4. LB winner would therefore need to win 5-6 games total, while WB winner needs 2-4. "True" extended series would make matches longer, not shorter. Whether that's a good thing or not is debateable, as is the logic that suggests its use. Current MLG rules are unequivocally unfair for WB winners, however. In the case of extended series in grand finals (thankfully rare), WB winners receive no advantage or are penalized for coming from winners.
This is so unclear. I honestly have no idea what you're trying to say, even after reading it multiple times.
Under current rules, if there's an extended series in the grand finals, there is only one Bo7 played. That means that, like Naniwa/Kiwi in Dallas, there is functionally no advantage for being the winner's bracket winner. He only receives the extended series advantage. So if A beat B 2-1, if they meet in loser's, then A needs 2 games and B needs 3, which is fair. However, if they meet in grand finals, A needs 2 games and B needs 3, and after that series the tournament is over. If they hadn't met beforehand, then A would need 2 and B would need at least 4. Thus, A at best breaks even, or is penalized a set for having met B before the grand finals.
If MLG wanted to be fair/consistent, then the first game should be an extended series, yes, but then, assuming A lost, there would have to be another one, since A still has a Bo3 to fall back on. Except they don't do that.
|
On August 23 2011 12:25 VikingKong wrote:
If they hadn't met beforehand, then A would need 2 and B would need at least 4. Thus, A at best breaks even, or is penalized a set for having met B before the grand finals.
This is incorrect. If the haven't met before, they would first play a bo3. If A wins that bo3, then A wins the tournament. If B wins, then the series would go to bo7 with B having the lead from the bo3.
|
You say "Current MLG rules are unequivocally unfair for WB winners, however.", but has anyone actually shown quantitatively that this is the case? Back in November, after MLG Dallas, someone made a statistical analysis comparing Single Elimination, Double Elimination, Double Elimination with Extended Series, and Round Robin formats. What he found, is that the Double Elimination with Extended Series slightly improves the outcomes of tournaments, but also does a good job of letting the better player move forward throughout the tournament, when compared with traditional Single and Double Elimination formats.
http://www.teamliquid.net/forum/viewmessage.php?topic_id=168345
It's a very interesting thread to read if you have time, and I have yet to see something similar showing that the extended series is a bad addition to the format. If people have shown that extended series is a worse format, then I'd be happy to see the analysis, and I'll retract my statement.
|
On August 23 2011 12:35 Azzur wrote:Show nested quote +On August 23 2011 12:25 VikingKong wrote:
If they hadn't met beforehand, then A would need 2 and B would need at least 4. Thus, A at best breaks even, or is penalized a set for having met B before the grand finals. This is incorrect. If the haven't met before, they would first play a bo3. If A wins that bo3, then A wins the tournament. If B wins, then the series would go to bo7 with B having the lead from the bo3.
Yeah, apologies for phrasing badly. A can win with 2 maps before the extended series, or win the extended series with 3-4 more maps. However, B requires 4 wins no matter what. He has to go at least 2-1, 4-3. Thus, B needs to win 4 games, whereas, had they met before, B would need to win either 3 or 4. If he needs to win 3, he "makes up" for it by having won 1 previously, but that doesn't change the fact that A does not receive an advantage for coming from winner's. B should have to win 2 Bo3 anyway. MLG's rule is assuming that their previous game is part of the grand finals, which is exceedingly odd because I can't think of a reason why they would include a previous match as part of the finals, and unfair because then A wants to face someone other than B, because if he won 2-1 he's actually being penalized rather than helped by the extended series rule.
On August 23 2011 12:36 Mr. Wiggles wrote:You say "Current MLG rules are unequivocally unfair for WB winners, however.", but has anyone actually shown quantitatively that this is the case? Back in November, after MLG Dallas, someone made a statistical analysis comparing Single Elimination, Double Elimination, Double Elimination with Extended Series, and Round Robin formats. What he found, is that the Double Elimination with Extended Series slightly improves the outcomes of tournaments, but also does a good job of letting the better player move forward throughout the tournament, when compared with traditional Single and Double Elimination formats. http://www.teamliquid.net/forum/viewmessage.php?topic_id=168345It's a very interesting thread to read if you have time, and I have yet to see something similar showing that the extended series is a bad addition to the format. If people have shown that extended series is a worse format, then I'd be happy to see the analysis, and I'll retract my statement.
I've just read the thread, and I personally agree with the concept of extended series. However it doesn't deal with the MLG-style rules in the grand finals, where the WB winner receives only the extended series advantage. I like the concept that no one should have a winning record against another player yet still lose (Nony/Tasteless example) and I'm happy that the data seems to support my intuition, but potentially penalizing the winner bracket finalist by treating their previous encounter as the first Bo3 in the final is unsettling to me, as I talked about in reply to Azzur. It suggests that a Bo3 played in pool play is the same as a Bo3 in the grand finals.
|
On August 23 2011 12:43 VikingKong wrote:Show nested quote +On August 23 2011 12:35 Azzur wrote:On August 23 2011 12:25 VikingKong wrote:
If they hadn't met beforehand, then A would need 2 and B would need at least 4. Thus, A at best breaks even, or is penalized a set for having met B before the grand finals. This is incorrect. If the haven't met before, they would first play a bo3. If A wins that bo3, then A wins the tournament. If B wins, then the series would go to bo7 with B having the lead from the bo3. Yeah, apologies for phrasing badly. A can win with 2 maps before the extended series, or win the extended series with 3-4 more maps. However, B requires 4 wins no matter what. He has to go at least 2-1, 4-3. Thus, B needs to win 4 games, whereas, had they met before, B would need to win either 3 or 4. If he needs to win 3, he "makes up" for it by having won 1 previously, but that doesn't change the fact that A does not receive an advantage for coming from winner's. B should have to win 2 Bo3 anyway. MLG's rule is assuming that their previous game is part of the grand finals, which is exceedingly odd because I can't think of a reason why they would include a previous match as part of the finals, and unfair because then A wants to face someone other than B, because if he won 2-1 he's actually being penalized rather than helped by the extended series rule.
I do understand what you're trying to say, but A is not without any advantage. Granted, the advantage A would've got in a traditional double-elim will be greater:
- If A wins the first bo3, then A wins the tournament.
- A has played less games and thus has more energy.
Anyways, it appears that you like the double-elim format. However, to make the finals in a double-elim completely fair, B needs to win 2 boXs. Many people have stated that this is a huge hurdle for the LB person to overcome and can lead to anti-climatic finals.
|
On August 23 2011 12:56 Azzur wrote:Show nested quote +On August 23 2011 12:43 VikingKong wrote:On August 23 2011 12:35 Azzur wrote:On August 23 2011 12:25 VikingKong wrote:
If they hadn't met beforehand, then A would need 2 and B would need at least 4. Thus, A at best breaks even, or is penalized a set for having met B before the grand finals. This is incorrect. If the haven't met before, they would first play a bo3. If A wins that bo3, then A wins the tournament. If B wins, then the series would go to bo7 with B having the lead from the bo3. Yeah, apologies for phrasing badly. A can win with 2 maps before the extended series, or win the extended series with 3-4 more maps. However, B requires 4 wins no matter what. He has to go at least 2-1, 4-3. Thus, B needs to win 4 games, whereas, had they met before, B would need to win either 3 or 4. If he needs to win 3, he "makes up" for it by having won 1 previously, but that doesn't change the fact that A does not receive an advantage for coming from winner's. B should have to win 2 Bo3 anyway. MLG's rule is assuming that their previous game is part of the grand finals, which is exceedingly odd because I can't think of a reason why they would include a previous match as part of the finals, and unfair because then A wants to face someone other than B, because if he won 2-1 he's actually being penalized rather than helped by the extended series rule. I do understand what you're trying to say, but A is not without any advantage. Granted, the advantage A would've got in a traditional double-elim will be greater: - If A wins the first bo3, then A wins the tournament. - A has played less games and thus has more energy. Anyways, it appears that you like the double-elim format. However, to make the finals in a double-elim completely fair, B needs to win 2 boXs. Many people have stated that this is a huge hurdle for the LB person to overcome and can lead to anti-climatic finals.
I definitely agree that A has a large advantage either way for both the reasons you bullet-pointed, but normally he would receive a map-score/series benefit for not having lost a match. In addition, it's a little unfair to have only the winner bracket winner punished like that. The other winner bracket finalist will have played only one BoX more than the winner, yet receives the map-score/series benefit. It also seems unfair/inconsistent to have extended series be a boon to the winner of the previous series in every case EXCEPT for grand finals, where the winner needs it most. It does, however, give the loser a better chance to come back, it is true, so it does build viewer excitement.
Yeah, I do, and I agree with needing 2 BoXs and the resultant detrimental effects on viewer excitement and the quality of the finals. Thus, like I said a couple of posts ago, I like Chill's idea of just having a double elim up until the Ro16 or so, and then playing single elim from there. It still solves the bad draws, but allows for more exciting matches. Yeah, it's a little arbitrary to just start single elim from Ro16 onwards, but seeding can be based upon their places on the double elim format so people still have a reason to stay in upper bracket. It's certainly better than starting single elim at the grand finals, imo, even if the WB winner still gets a 1/2 map advantage in a Bo5/7.
Hahah, this blog derailed fast  I was just reading that MLG Q&A thread when I just had to rant about people who couldn't tell between extended series and normal double elim. Now this blog is all serious debating about double elim and extended series.
|
My bad! Accidentally posted instead of edited.
|
This is an addendum to what Mr. Wiggles was saying above: a quick quantitative analysis comparing the double-Bo3 format for grand finals to the extended series rule. It's a bit more rough around the edges than the analysis Mr. Wiggles links to, but perhaps easier to see through the logic. The conclusion is, roughly speaking, that the extended series rule is better for the WB player than the double-Bo3 rule in the case that the original series was 2-0, and worse for the WB player than the double-Bo3 rule if the original series was 2-1.
+ Show Spoiler [analysis] +
I'll be doing the analysis from the WB player's point of view, so "W" means the WB player won, "L" means he lost. It's simple enough to write down all possibilities for a sequence of six games in a row (the most that will be played under any format), and write down in each case whether the WB player is awarded the win for the finals or not. For instance, the sequence of games LLWLWW gives the win to the WB player if the rule is double Bo3 or the first extended series went 2-0, but gives him a loss if the first extended series went 2-1.
If we make the assumption that the two players are equally skilled then each such string of 6 games is equally likely. (Otherwise, one weights by the difference in skill as appropriate, but if you want to be this complicated you may as well look at the analysis linked to above.) Below are listed the possibilities on which the double-Bo3, extended 2-0, and extended 2-1 possibilities disagree. Out of the 64 different possibilities, there are only 10 on which the rules ever disagree (meaning this whole discussion should only matter about 16% of the time):
Six possibilities win the double Bo3 and the extended 2-0, but lose the extended 2-1:
LLW LWW
LLW LWL
LLL WWW
LLL WWL
LWL LWW
WLL LWW
Two possibilities lose the double Bo3 but win either extended series:
LWL WLL
WLL WLL
Two possibilities win the 2-0 extended series, but lose the 2-1 and the double Bo3:
LWL LWL
WLL LWL
In total, out of these 10, all win the extended 2-0, six win the double Bo3, and two win the extended 2-1. The extended series rule gives an advantage to the WB player (versus the double Bo3) if he wins the first series 2-0, and it gives an exactly corresponding disadvantage if he wins the first series 2-1.
So before an MLG grand finals that will be an extended series (when you already know the results of the previous series), you can say to yourself (assuming the players are roughly equally matched!) "there's a 1 in 16 chance that the extended series rule, versus the double Bo3 rule, will matter this time".
I make no claim that this analysis settles whether the rule is a good one, but it does suggest at least that it's not at all clear whether current MLG rules are unfair for WB winners.
Also of note is that this analysis does not apply to the effects of the extended series rule where two people meet in the losers bracket. There it does (clearly) give an advantage to the player who won the first series compared to a more traditional double-elimination format. Again, whether this is desirable is unclear to me.
|
On August 23 2011 13:16 incnone wrote:This is an addendum to what Mr. Wiggles was saying above: a quick quantitative analysis comparing the double-Bo3 format for grand finals to the extended series rule. It's a bit more rough around the edges than the analysis Mr. Wiggles links to, but perhaps easier to see through the logic. The conclusion is, roughly speaking, that the extended series rule is better for the WB player than the double-Bo3 rule in the case that the original series was 2-0, and worse for the WB player than the double-Bo3 rule if the original series was 2-1. + Show Spoiler [analysis] +
I'll be doing the analysis from the WB player's point of view, so "W" means the WB player won, "L" means he lost. It's simple enough to write down all possibilities for a sequence of six games in a row (the most that will be played under any format), and write down in each case whether the WB player is awarded the win for the finals or not. For instance, the sequence of games LLWLWW gives the win to the WB player if the rule is double Bo3 or the first extended series went 2-0, but gives him a loss if the first extended series went 2-1.
If we make the assumption that the two players are equally skilled then each such string of 6 games is equally likely. (Otherwise, one weights by the difference in skill as appropriate, but if you want to be this complicated you may as well look at the analysis linked to above.) Below are listed the possibilities on which the double-Bo3, extended 2-0, and extended 2-1 possibilities disagree. Out of the 64 different possibilities, there are only 10 on which the rules ever disagree (meaning this whole discussion should only matter about 16% of the time):
Six possibilities win the double Bo3 and the extended 2-0, but lose the extended 2-1:
LLW LWW
LLW LWL
LLL WWW
LLL WWL
LWL LWW
WLL LWW
Two possibilities lose the double Bo3 but win either extended series:
LWL WLL
WLL WLL
Two possibilities win the 2-0 extended series, but lose the 2-1 and the double Bo3:
LWL LWL
WLL LWL
In total, out of these 10, all win the extended 2-0, six win the double Bo3, and two win the extended 2-1. The extended series rule gives an advantage to the WB player (versus the double Bo3) if he wins the first series 2-0, and it gives an exactly corresponding disadvantage if he wins the first series 2-1.
I make no claim that this analysis settles whether the rule is a good one, but it does suggest at least that it's not at all clear whether current MLG rules are unfair for WB winners.
Also of note is that this analysis does not apply to the effects of the extended series rule where two people meet in the losers bracket. There it does (clearly) give an advantage to the player who won the first series compared to a more traditional double-elimination format. Again, whether this is desirable is unclear to me.
Nice analysis! I roughly ran through something along those lines mid-way through my rant just to make sure I wasn't talking out of my ass. It just seems odd to me that extended series is always an advantage in loser's, and yet not in the grand finals. Also, it's hard to compare how a player plays in group/loser and in grand finals. Some players might have nerve issues in the finals, so MLG's rules would help, but others might react more positively to stress. Builds might also be hidden for finals, which would make it even harder to guess at the total effects since all of those apply to both sides, though I would go with my gut in saying that most people would do better in the finals than in group, and that the better player would be the player doing the best in the worst conditions, ie. the finals.
|
|
|
|
|
|
EPT
