|
MLG's Extended Series Crushing Dreams
MLG Spring Arena 2 wrapped up this weekend, and like every tournament Spring Arena has its stories — like Queens and Polt’s TvZ. A story most will miss about this weekend is about a young, unknown player from team MvP, Dream. Dream finished 8th and flew home with a cheque for $900 and a plane ticket for another chance at the MLG Spring Championship in June. All things considered, not a bad deal. But it could have been more.
It’s often been that international tournaments, run over a short time span, provide an opportunity for the breakout performance of a capable player yet to prove themselves on the long, tough road of GSL. PuMa, MMA, and DongRaeGu — three of the biggest earners in the game — all made big wins overseas before making a single appearance in Code S, the most prestigious stage in South Korea. Symbol, another player from this weekend, could be added to that list on the strength of his results in May alone when he won’t debut in Code S for another few weeks. But we won’t be adding Dream, at least not yet. Overshadowed by the performances of other rising stars like Symbol, viOLet, and Heart, Dream still has more work to do.
It’s easy to assume that Dream just isn’t quite good enough. After all, if he was, why didn’t he place higher? Why didn’t he win? This would be a reasonable argument if Dream had a fair shake. He didn’t. MLG’s extended series rule isn’t meant to be unfair. It wasn’t implemented to punish some players arbitrarily and not others. But, meaning well doesn’t stop poor rules from being poor, and this time Dream drew the short end of the stick. Every player at MLG Spring Arena 2 had two chances. MLG only gave Dream one.
Extended Series (ES) is a pretty simple concept. When a rematch occurs in a tournament, ES dictates that the match continue the score from the original series in a BO7 rather than starting a new BO3. When Heart met HuK in the final four of MLG Winter Championship 2012, they had already squared off earlier that weekend in Group C, with Heart winning (2-1). In the rematch to decide who would move on to face DongRaeGu, HuK needed to win an additional three games while Heart only needed to win two.[1] ES not only rewards Heart for his victory over HuK earlier in the tournament, it gives a stronger reward if the victory is more convincing (2-0, rather than 2-1). Nevertheless, ES is almost unanimously criticized. Not a single MLG event passes without the Starcraft community, both players and fans, throwing their arms up in exasperation and cursing whoever might have enacted this heinous rule and/or works to keep it in tact.[2]
The principle criticism for ES lies in that the format for a traditional double elimination tournament[3] already penalizes players in the losers bracket by moving them through the tournament at a slower rate than those in the winners bracket — the earlier you lose your first game, the more games you’ll have to win overall to reach the same ranking.[4] Furthermore, if you ultimately make the finals, you must win twice. When Dream and Symbol faced off for the second time, Symbol had won one match and lost one match since their first meeting; Dream had won four straight. After losing, Symbol was relegated to the losers bracket where he would meet Dream again. At that point, the tournament system considers them equals — but MLG doesn’t. MLG saddles Dream with a handicap, a handicap he wouldn’t have if the bracket draw gave him a different opponent instead.
The Upside of ES
MLG is as much, if not more, about spectators than it is about players. Spectating a tournament is exciting because of what is on the line in each and every match — when mainstream professional sports enter playoff season, television ratings skyrocket in response. These tournaments are single-elimination brackets where each victory is crucial because once a match is lost, the season is over. There are no second chances. In tournaments that use different structures, like MLG, this isn’t the case. A player can lose a match and then continue on in the tournament , and potentially rematch the same opponent later on. When this happens, it is often disappointing. We already have a result that we assume carries weight. In MLG, which takes place over a weekend, this result is still warm! Just or unjust, playing again a day or even a few hours after a previous bout is an awful situation. If the second series goes the same as the first, it looks like a waste of time. If the tables turn and the other player emerges victorious, one player is eliminated over what looks like a tie.[5] ES attempts to sidestep this conundrum by offering one definitive result rather than two that could disagree, and allowing for a close series to last longer than a blowout.
Though they are a major player in the Starcraft 2 circuit, MLG was not built on running tournaments for Starcraft or Brood War. MLG has its roots in console shooters (FPS), and then fighters (FG), and then Starcraft. These games are all different from each other. The men behind the scenes at MLG are passionate and experienced in gaming, but not necessarily in Starcraft. These differences are important to the topic at hand because I believe ES has a much stronger effect on the tournament outcome in Starcraft than it does in other games, turning ES from a well intentioned quirk of the tournament into a potential disaster for a random, unlucky player.
How ES and Starcraft Don’t Mix
Strategical planning in Starcraft along with imperfect information is a large part of what makes the game what it is. A player formulates a general plan from a wide variety of options most appropriate to the situation (matchup, map, etc.), makes adjustments based on information about his opponent’s plan that he learns through the game, and attempts to conceal information about his own plans. It’s not unusual to see the same player against the same opponent utilize three completely different approaches throughout a series. At their core, FG and FPS[6] do not have this in common with Starcraft. FPS are not perfect information, but neither the hidden information (location, weapons, etc.) or methods of detection/concealment (sound, being shot) are nearly as complicated as they are in Starcraft. Strategically, games are also much more simplistic. A high level FPS game between three different pairs of players will look quite similar, as all players have the same options available and preference for those options will not vary much — the game is contested based on tactics and execution in a dynamic, fast-paced environment. The strategy involved in the game is more like a hierarchy than a palette — a strategy to control a map or situation may be better than another, but there is not much element of choosing between many similarly effective options and little in the way of detecting your opponent’s strategy. FG are perfect information — nothing is hidden (health, energy, etc.) between you and your opponent. FG do have a diverse array of characters, creating an aspect of strategical planning similar to the different matchups in Starcraft.[7] A player must not only identify the strengths and attributes of their character, but also how they relate to the other characters their opponents may choose. Since the game is perfect information, however, concealment is not a factor. A player knows exactly what his opponent is capable of at any particular moment. This is in contrast to Starcraft, where its possible one player can accumulate a strong material advantage in secret.
Lacking a complex strategical element do not make these games less entertaining to watch or play, but simply makes them different. There are other concepts at work in either genre which Starcraft has little or none of. But I think this particular aspect of strategical planning in Starcraft has a peculiar effect on how games play out between two specific players. Being down in a series has a tangible effect on how desirable the various options available are, something beyond the psychological pressure of being behind on points. This simply does not have a place in FPS or FG the way it does in Starcraft — the games are just different.
Many sports and games more or less follow a linear hiearchy. A>B>C, ergo, A>C. ES puts a lot of faith into this idea. It assumes that if one player beats another, he is not just superior but entirely superior. Should a rematch occur, the previous loser needs to prove himself beyond a shadow of a doubt. Starcraft results, unfortunately, do not correlate well at all with a linear hierarchy. There is a much looser correlation between defeating a player and being more capable than that player to win against the next opponent due to considerations like matchup and styles. As spectators we accept or ignore this fact because of just how exciting a tournament is and trust that things will eventually work out, but it is intuitively understood by fans of the game. When two players meet for a second time before it is absolutely necessary, it comes across as disappointing and unfortunate. If necessary, it can seem redundant if a strong case is not made for why they must play again.[8] ES tries to act like a band-aid, but only makes a bad situation worse with its misguided intentions.
For these reasons and others, Starcraft has a lot of variance. Between two top competitors, a clearly superior performance will include a surprisingly high number of losses.[9] This makes overcoming a handicap like ES much more difficult than it seems. If two players are evenly matched, the influence of the advantage in an ES can equal or even exceed the influence of the skill difference between the players.[10] If this is not the case and one player is much better than the other, ES has little effect. Unfortunately, this means that the more potential to be exciting a game has, the more ES ruins the party. It does so in the name of maintaining the integrity of the first series, but in doing so sullies the second. The second series will always be later in the tournament — more viewers, and more significance. Which is more important?
The Solution
A tournament is a complex system, full of decisions with all sorts of consequences. It’s erroneous to assume there is one perfect structure for all situations, or that the answer to a problem is as simple as “extended series sucks, remove it”. When a match takes place in the finals of a double elimination tournament, ES functions much better. One player receiving a disadvantage is part and parcel of the format[11], and running one long series is much more sensible than running two series if the players had not met before. It’s also sensible if, as is often likely with double elimination tournaments, the players had recently played in the winners bracket. The non-linear hierarchy does not come into play because there are no other players left in contention — in the finals, the only thing that matters is who wins between the last two players.
But with more players left in the tournament, rematches are the true problem and ES is not an adequate solution. The only thing to do is to look at the entire tournament structure with the goal of avoiding rematches entirely or minimizing them along with providing a different context whenever rematches are possible. Dream and Symbol didn’t have to rematch in the tournament, not that early. MC and Inori were both in the same round, and no other combination of the four players had occurred yet that weekend. A rematch between the two players could have been avoided until two rounds later. Every player save the champion at MLG this past weekend lost against two opponents — but Dream just lost to Symbol. Forced to avenge his loss from a 0-2 deficit, his second try wasn’t even a legitimate chance. We'll see Dream again in Anaheim in a couple weeks. Hopefully, MLG won't cripple him this time.
Notes
1. Don’t fret, HuK fans. ES was on his side just a few rounds earlier against Socke.^
2. He doesn’t seem to be hard to find.^
MLG’s championship tournaments use a modification of double elimination which segregates some players for group play before adding them into the later stages of the tournament. Winter Arena and Spring Arena 2 were both classic double elimination.^
4. Rule of thumb is that every win in the winners bracket starting in the second round is worth two wins in the losers bracket. There is no difference in total number of games played if the initial loss is in the first or second round.^
5. When splitting two isolated series, it’s possible for one player to actually come out ahead in games overall and still be eliminated in a typical double elimination format. In two BO3, for example, Player A wins the first series 2-0 and then loses the second series 1-2. This makes him 3-2 overall, but eliminated.^
6. Unlike deathmatch based FPS, these two concepts have a stronger presence in team/mission based FPS such as CounterStrike.^
7. FG where one chooses a combination of characters, such as in the Marvel vs. Capcom series, adds to this strategical complexity.^
8. In GSL Code S Season 1, Genius and DongRaeGu found themselves in the same group during both stages and then met again in the finals. This wasn’t a problem because there is a clear difference, not to mention quite a bit of time, between a match in the first group stage and a BO7 series in the finals of the entire tournament.^
9. Using TLPD data in SC2 and BW suggests that dominant win rates in non-open leagues and tournaments (like GSL or OSL) are rarely above 70%. This data is not totally reliable for several reasons. One reason, for example, is that TLPD counts game win rates and not match win rates, and it’s perfectly reasonable that a player might maximize his match winning percentage by increasing his game winning percentage. Even if we had a more reliable number, a series of matches between two players would be constantly fluctuating as the players learn, adapt, etc. Regardless, I think 65-75% is a decent estimate for these purposes.^
10. Let’s assume that Player A will win 60% of games and Player B will win 40%. Player A will win 64.8% of BO3 matches. If he’s playing an extended series down 2-1, though, Player A will lose 66.304% of BO7s. This means he is more likely to lose from the disadvantage in an extended series than he would be to win a straight BO3 — the single game disadvantage is stronger than the difference between them in skill. Again, win rates are subject to the same lack of reliability outlined in the previous footnote.^
11. Whether starting the finals with a handicap is a good thing is another question all together. My answer is absolutely not, but I’ll cover that in a future article.^
 
|
this is extraordinarily well-written 5/5
|
On May 25 2012 06:07 mockturtle wrote:How ES and Starcraft Don’t MixStrategical planning in Starcraft along with imperfect information is a large part of what makes the game what it is. A player formulates a general plan from a wide variety of options most appropriate to the situation (matchup, map, etc.), makes adjustments based on information about his opponent’s plan that he learns through the game, and attempts to conceal information about his own plans. It’s not unusual to see the same player against the same opponent utilize three completely different approaches throughout a series. At their core, FG and FPS [6] do not have this in common with Starcraft. FPS are not perfect information, but neither the hidden information (location, weapons, etc.) or methods of detection/concealment (sound, being shot) are nearly as complicated as they are in Starcraft. Strategically, games are also much more simplistic. A high level FPS game between three different pairs of players will look quite similar, as all players have the same options available and preference for those options will not vary much — the game is contested based on tactics and execution in a dynamic, fast-paced environment. The strategy involved in the game is more like a hierarchy than a palette — a strategy to control a map or situation may be better than another, but there is not much element of choosing between many similarly effective options and little in the way of detecting your opponent’s strategy. FG are perfect information — nothing is hidden (health, energy, etc.) between you and your opponent. FG do have a diverse array of characters, creating an aspect of strategical planning similar to the different matchups in Starcraft. [7] A player must not only identify the strengths and attributes of their character, but also how they relate to the other characters their opponents may choose. Since the game is perfect information, however, concealment is not a factor. A player knows exactly what his opponent is capable of at any particular moment. This is in contrast to Starcraft, where its possible one player can accumulate a strong material advantage in secret. Lacking a complex strategical element do not make these games less entertaining to watch or play, but simply makes them different. There are other concepts at work in either genre which Starcraft has little or none of. But I think this particular aspect of strategical planning in Starcraft has a peculiar effect on how games play out between two specific players. Being down in a series has a tangible effect on how desirable the various options available are, something beyond the psychological pressure of being behind on points. This simply does not have a place in FPS or FG the way it does in Starcraft — the games are just different.
This paragraph is pretty much total bullshit; please learn about FPS games and fighting games before writing about them.
Particularly, fighting games aren't perfect information in practice because so many things happen faster than human reaction time (roughly 20 frames if you're looking specifically for an option, but if you're not looking for one specific thing, closer to 24 frames), and detecting and countering your opponent's strategy is the core of almost any competitive FPS.
Many sports and games more or less follow a linear hiearchy. A>B>C, ergo, A>C. ES puts a lot of faith into this idea. It assumes that if one player beats another, he is not just superior but entirely superior. Should a rematch occur, the previous loser needs to prove himself beyond a shadow of a doubt. Starcraft results, unfortunately, do not correlate well at all with a linear hierarchy. There is a much looser correlation between defeating a player and being more capable than that player to win against the next opponent due to considerations like matchup and styles. As spectators we accept or ignore this fact because of just how exciting a tournament is and trust that things will eventually work out, but it is intuitively understood by fans of the game. When two players meet for a second time before it is absolutely necessary, it comes across as disappointing and unfortunate. If necessary, it can seem redundant if a strong case is not made for why they must play again.[8] ES tries to act like a band-aid, but only makes a bad situation worse with its misguided intentions.
No one has ever made that assumption about a fighting game or an FPS. Well, at least not anyone who actually knows what they're talking about.
|
This was well written but it doesn't really bring up anything that hasn't been brought up before the many times ES have been discussed. In the end it comes down to a different viewpoint/goal in regards to what MLG want their tournament to do (at least according to MLG).
MLG want the tournament to be as definitive as possible within time constraints to determine the best player at the tournament. When you ask who is the better player if they play 2 Bo3 against each other a lot of the time you won't get a very good answer because they will tie. In the case of double elim a player with less wins can progress past the person that beat them more overall games. When you ask who is the better player in a Bo7 you get a decent answer. That's really the sticking point here, MLG cares about a double-elim format because it will usually give you a better idea of who the better players are. In their eyes that gets undercut when a player advances over another that they have less games against so they put in ES.
I myself don't have a strong opinion either way though I would prefer a regular double-elim format just because it's easier for people to understand.
|
On May 25 2012 07:24 Obscura.304 wrote:
This paragraph is pretty much total bullshit; please learn about FPS games and fighting games before writing about them.
Particularly, fighting games aren't perfect information in practice because so many things happen faster than human reaction time (roughly 20 frames if you're looking specifically for an option, but if you're not looking for one specific thing, closer to 24 frames), and detecting and countering your opponent's strategy is the core of almost any competitive FPS.
I did the best to challenge my hypothesis before publishing it but there isn't a whole lot of information out there (like say, the equivalent of liquipedia) I could use to assess them. If I knew more about FPS and fighting games (I have a little experience), it would mean I knew less about RTS.
Being fast does not make something "not perfect information". It's perfect information. You know what your opponent has (e.g. health), and you know what moves his character has available. Contrast this to SC2 where you never get full vision of what exactly your opponent is doing -- he could take an expansion, add gateways, change tech, whatever and you don't know as soon as it happens. You make predictions, guesses, adjustments, and part of those adjustments are to continue gaining more information (so you can make more adjustments, etc).
Bear in mind, perfect information is a term from game theory. Game theory is not video game design, it's a branch of mathematics. It's meant to describe a game like chess compared to a game like poker. These are turn based games, not real time. Making something real time is always going to add some measure of reactivity and perception and this is going to change from game to game. I'm calling fighters perfect information because there is no equivalent of "scouting" or denying scouting like we have in RTS, nor is there information that's privy to one player but not the other (other than his own thoughts). FPS like quake adds some of these elements but it's kinda simple -- your opponent doesn't know,say, how much ammo you have or what weapons you have (he can guess once you shoot him with it) or how much health he has (he can guess based on your exchanges). Is perfect/imperfect information the right term for a real time game? Maybe not, maybe it should have its own terminology... but afaik that doesn't exist. If you want to argue that hidden vs. non hidden information is meaningless when a game is in real time, that's another discussion.
Also, because fighters are perfect information (by my definition), there is much more complicated decision making and prediction that takes place at a super fast pace... the game becomes more intricate and complex in different ways than it does in Starcraft. You don't need to actively "learn" about your opponent, but there is much more pressure to, say, make really fast predictions about a small movement your opponent will do next. A comparison in SC I could think is choosing where to storm a bio force -- I look at the positioning of my army, his army, and try to figure out where is the best place to aim a psi storm. Maybe I'm not trying to win the battle but actually delay it, so I'd storm in a different place, once that doesn't maximize damage to his units but forces him to retreat a little. I'd suggest that this is a similar KIND of decision making that takes place in fighting games (aiming the psi storm against marines which are dancing and kiting playing the role of the short term prediction), only in a fighter these thoughts and actions are much, much more complicated.
I didn't mean in any way to suggest that fighters or fps games are stupid, for idiots, or not fun. I don't think anyone is going to take issue with my suggesting that the games are different. One of SC's biggest weaknesses as a spectator game, IMO, is that the player who is losing is at a much larger disadvantage... you could say the winning player accumulates momentum exponentially rather than linearly. This makes for games which can be rather boring -- you know who is going to win, but it will take a while to get there. Other genres are more ripe for comebacks. This weakness also makes ES more unjust in my opinion.
No one has ever made that assumption about a fighting game or an FPS. Well, at least not anyone who actually knows what they're talking about.
Everyone has made that assumption about every game or sport, including SC. I think its usefulness is sort of a line of grey -- in sprinting, this holds pretty strongly, we would only see it proven wrong when a single persons time varies significantly from one race to the next. In tennis we have different surfaces and a lot of data to show that they change the game, so however useful A>B>C is at telling us A>C, we know it's much less useful if we move A>C to a new surface.
Ignoring maps and styles, each player is subject to three matchups in an SC tournament which creates a really wacky relationship between the players since they all will have a different strength in each matchup (give a player with strong ZvZ only Z opponents and he'll go far, no matter how bad his ZvP or ZvT is). I can't begin to guess whether this is more or less significant than it would be in fighting games (different characters instead of different races), though I'd love it if there was a fighter liquipedia that I could sink my teeth into. I'd guess it's slightly less of an effect in an death match FPS since you don't have the equivalent of racial matchups. But regardless, I'm not trying to assert in that section that FPS or FG are directly linear and that SC isn't, simply that SC is very non-linear.
Regardless, if after clarifying my position you still think I am way off the mark, I'd welcome you to elaborate further either here or in private and help me to find examples (whether it's data from a tournament or videos from matches). I'm far more attached to the idea that games are different and tournaments need to be structured with the nuances of the game (something I have not seen people make anywhere) in mind than I am to any specific assertion I'd make about one game or another. This goes out to anyone else who has a lot of experience/passion for a fighter or an fps. This general topic is really fuzzy, there's not much to do other than to get a bunch of experienced opinions on the subject and try to create a common language.
|
On May 25 2012 07:34 GogoKodo wrote:
When you ask who is the better player in a Bo7 you get a decent answer. That's really the sticking point here,
My assertion is twofold,
1) you don't get a decent answer -- the cards are strongly stacked in favor of the person who won the first series (even if the "worse player" wins the first series). If we were extending a BO3 to like a BO15, then the small advantage you get from a BO3 is totally nullified in the face of a massive number of games. But going from a BO3 to a BO7 means that a one game advantage is pretty big, let alone two.
2) if your definition of best player is not who wins between two players but who is most capable against the remaining players in the tournament, the outcome of two closely matched players is not super-relevant either...
In my opinion, anytime there's a situation where two players are close it's really not a catastrophe either way... so the underdog wins, no big deal, he's almost as good a player anyway. On the grand scheme of things however, forcing someone into a situation where he's at an extreme disadvantage (such as at spring arena, where Dream had to face the same opponent from a disadvantage rather than getting a fresh opponent from scratch) IS a catastrophe.
|
and you know what moves his character has available.
And, see, if you knew anything about fighting games, you'd know this was false.
Let's say you're playing BlazBlue, your Jin vs. an opponent's Ragna. Ragna does his Inferno Divider, and has 50% meter; you block it.
In the 10 or so frames after you block the move, what moves can Ragna do? Answer: you don't know, because there's no way to know that quickly whether he Rapid Canceled the move or not. If he did not RC the move, then he cannot do any move at all; however, if he did RC the move, then he has several options open to him.
Another example of this is characters with "stored supers"; in Super Street Fighter 2: Turbo, if Chun-Li has 100% meter, there is no way of knowing whether she has her super "stored" or not if she's walking forward towards you.
Also, because fighters are perfect information (by my definition), there is much more complicated decision making and prediction that takes place at a super fast pace... the game becomes more intricate and complex in different ways than it does in Starcraft. You don't need to actively "learn" about your opponent,
Sorry, but that's basically the exact opposite of true. To play fighting games worth a damn, you have to know all about your opponent's tendencies (and, when playing against an unknown opponent, you have to figure them out on the fly).
One of SC's biggest weaknesses as a spectator game, IMO, is that the player who is losing is at a much larger disadvantage... you could say the winning player accumulates momentum exponentially rather than linearly. This makes for games which can be rather boring -- you know who is going to win, but it will take a while to get there. Other genres are more ripe for comebacks. This weakness also makes ES more unjust in my opinion.
But this aspect is also true in fighting games. In most FGs, the winning player gains meter faster than the losing player, and in "team" games like the MvC series, having fewer characters left on your team is a *massive* disadvantage, since you have less damage output and you lack assists.
. I can't begin to guess whether this is more or less significant than it would be in fighting games (different characters instead of different races), though I'd love it if there was a fighter liquipedia that I could sink my teeth into.
shoryuken.com has a similar wiki.
I'd guess it's slightly less of an effect in an death match FPS since you don't have the equivalent of racial matchups. But regardless, I'm not trying to assert in that section that FPS or FG are directly linear and that SC isn't, simply that SC is very non-linear.
Character matchups in FGs are actually a much bigger deal than racial matchups in an RTS. A game can have lots of 60-40 matchups and be considered *amazingly* balanced (hell, GGXX:AC is considered the best balanced FG of all time, and has a few matches that are sometimes claimed to be 70-30!). The most successful competitive FG of all time, Super Street Fighter 2: Turbo has matchups that are literally 90-10 (admittedly, most ST players know multiple characters; the same is not the case for GGXX:AC, though). This game succeeded as a tournament game for decades.
|
On May 25 2012 09:38 Obscura.304 wrote:
And, see, if you knew anything about fighting games, you'd know this was false.
Let's say you're playing BlazBlue, your Jin vs. an opponent's Ragna. Ragna does his Inferno Divider, and has 50% meter; you block it.
In the 10 or so frames after you block the move, what moves can Ragna do? Answer: you don't know, because there's no way to know that quickly whether he Rapid Canceled the move or not. If he did not RC the move, then he cannot do any move at all; however, if he did RC the move, then he has several options open to him.
Something happening faster than possible perception is better categorized as prediction of the future rather than hidden information. From your example, you can back up a step and simply include that rapid cancelling (and subequent moves) were part of the decision tree you'd need to sort through all at once (since the game is too fast for you to make multiple decisions)...
Since fighters are played the same with two players on one screen or two players on separate screens, there's no hidden information about each character. Intentions of each player, on the other hand, is a whole different ballgame. what's going on in your opponent's head is always hidden.
Another example of this is characters with "stored supers"; in Super Street Fighter 2: Turbo, if Chun-Li has 100% meter, there is no way of knowing whether she has her super "stored" or not if she's walking forward towards you.
I don't know this game & specific example to comment on it, however as you describe it it does sound accurate (as in, what I would qualify as imperfect information). These genres have broad general concepts but it's fully possible to borrow or translate aspects from one to another. For example, money in CS is a very "rts-y" concept -- you win small battles, gain more money than your opponent, and spend that money on stuff which makes it easier to win the next battle. DOTA games are a relatively new genre and it's pretty clear to see how aspects of it were taken from all kinds of other games.
Sorry, but that's basically the exact opposite of true. To play fighting games worth a damn, you have to know all about your opponent's tendencies (and, when playing against an unknown opponent, you have to figure them out on the fly).
this is not what I meant. figuring out your opponents style in a fighting game has a pretty strong parallel in sc2 where player style also has a role (econ oriented, allin oriented, etc). but there isn't a parallel in fighting games for scouting (i send a probe to his base, i see his buildings, i get specific indisputable data). now, do stylistic tendencies and feints play a stronger role in FG (or some FG) than they do in sc? another question all together.
But this aspect is also true in fighting games. In most FGs, the winning player gains meter faster than the losing player, and in "team" games like the MvC series, having fewer characters left on your team is a *massive* disadvantage, since you have less damage output and you lack assists.
this is 100% the same thing as what i was talking about, you're not wrong at all. in fact, i've had this exact conversation in a game design context. as far as severity... i don't know. things like meter or having characters on your team are clearly much more significant than simply having less health (something linear). i think the popularity of these concepts in games indicate that going beyond linear is good but still there's the question of how much. but how these accumulated advantages scale in real situations and then how to compare that them to starcraft requires someone quite versed in both games to draw comparisons or maybe a ton of data that doesn't exist.
.shoryuken.com has a similar wiki.
I meant as far as liquipedia's results, not about the game. looking at results (if they are detailed enough) one can draw certain conclusions, such as how often lower seeded players beat higher seeded players, how often a series goes 2-1 rather than 2-0, stuff like that. for example, you can pretty easily use TLPD and data from chess tournaments to quickly surmise that ELO is much more reliable in chess for predicting games than it is in SC2 without having any clue about either game.
Character matchups in FGs are actually a much bigger deal than racial matchups in an RTS. A game can have lots of 60-40 matchups and be considered *amazingly* balanced (hell, GGXX:AC is considered the best balanced FG of all time, and has a few matches that are sometimes claimed to be 70-30!). The most successful competitive FG of all time, Super Street Fighter 2: Turbo has matchups that are literally 90-10 (admittedly, most ST players know multiple characters; the same is not the case for GGXX:AC, though). This game succeeded as a tournament game for decades.
Players potential (separate from statistical potential) is an important factor. If, say (and yes, I am pulling these numbers out of my ass using actual character names from street fighter but they don't need to be accurate to make the example), Ryu is 60% against chun-li but 40% against Zangief and these results more or less hold true for *all* Ryu players (they do slightly better against chun li and slightly worse against zangief), then that presents one sort of problem. If specific players of Ryu are all over the place (some for example are 20% against chun li and 90% against zangief; others are average, others are opposite) that's a whole new type of problem similar to what goes on in SC2. Some players are much better at one matchup than the others or much worse at one matchup than the others and their success in a tournament varies wildly by what races they are matched up against. But, it was my estimation that huge variance matchup to matchup (outside of game balance) was not as much of a factor. This is actually something that we could get a pretty decent idea of with enough win ratio data.
I definitely would not begin to dispute that character choices function very similar to racial choices in both games (and even though a single character is less complex an entity than a single race, having a bunch of characters vs. three races steps that up), and if someone argued that your opponents character in your bracket draw was a more significant factor in FG than it was in SC I would be wide open to being proved wrong.
Regardless, I don't think ES is really a good idea for any game (if game results are so consistent that starting from a disadvantage is not crippling, than a plain BO3 should be reliable enough) -- simply that it's extra bad for SC. Maybe it's extra-bad for SC and FG equally. I definitely DO think that in a death match style FPS like quake that ES would play less of a role. This entire section of contrasting games was mostly me trying to figure out why MLG would think ES is a good idea in the face of constant complaints (like, say, they had a background in other games and failed to realize SC might work differently), and highlighting that different games require different tournament structures (for example, I'm passionate about the idea that SC tournaments should force diverse matchups early in a tournament -- the same logic could apply to FG, but be a bigger deal because all the characters constitute more diversity to cover).
|
On May 25 2012 10:18 mockturtle wrote:
Something happening faster than possible perception is better categorized as prediction of the future rather than hidden information. From your example, you can back up a step and simply include that rapid cancelling (and subequent moves) were part of the decision tree you'd need to sort through all at once (since the game is too fast for you to make multiple decisions)...
In Starcraft, we can take a step back and say "ok, at 6 minutes into the game, Terran can be doing these things...".
Another example of this is characters with "stored supers"; in Super Street Fighter 2: Turbo, if Chun-Li has 100% meter, there is no way of knowing whether she has her super "stored" or not if she's walking forward towards you.
I don't know this game & specific example to comment on it, however as you describe it it does sound accurate (as in, what I would qualify as imperfect information). These genres have broad general concepts but it's fully possible to borrow or translate aspects from one to another. For example, money in CS is a very "rts-y" concept -- you win small battles, gain more money than your opponent, and spend that money on stuff which makes it easier to win the next battle. DOTA games are a relatively new genre and it's pretty clear to see how aspects of it were taken from all kinds of other games[/quote]
Another example of imperfect information, this one a bit less esoteric, is Balrog's turn-around punch. At any given point in time, Balrog can deny himself access to two of his moves (by holding the buttons for two particular normals down); however, in return for this, he has access to a punch that can go through fireballs put out by zoning characters. If you're, say, Ryu facing a Balrog that is walking towards you, you do not know which moves he has access to; he could have access to all of his normals, or he could be sacrificing his footsies game to have that turn-around punch to get through a fireball.
this is 100% the same thing as what i was talking about, you're not wrong at all. in fact, i've had this exact conversation in a game design context. as far as severity... i don't know. things like meter or having characters on your team are clearly much more significant than simply having less health (something linear). i think the popularity of these concepts in games indicate that going beyond linear is good but still there's the question of how much. but how these accumulated advantages scale in real situations and then how to compare that them to starcraft requires someone quite versed in both games to draw comparisons or maybe a ton of data that doesn't exist.
There's a reason why J.Wong's MvC 2 "1 character vs 3 comeback" match is super-famous. Comebacks in MvC 2 at high level are almost completely unheard of.
Players potential (separate from statistical potential) is an important factor. If, say (and yes, I am pulling these numbers out of my ass using actual character names from street fighter but they don't need to be accurate to make the example), Ryu is 60% against chun-li but 40% against Zangief and these results more or less hold true for *all* Ryu players (they do slightly better against chun li and slightly worse against zangief), then that presents one sort of problem. If specific players of Ryu are all over the place (some for example are 20% against chun li and 90% against zangief; others are average, others are opposite) that's a whole new type of problem similar to what goes on in SC2. Some players are much better at one matchup than the others or much worse at one matchup than the others and their success in a tournament varies wildly by what races they are matched up against. But, it was my estimation that huge variance matchup to matchup (outside of game balance) was not as much of a factor. This is actually something that we could get a pretty decent idea of with enough win ratio data.
You're wrong on that estimation. For instance, in BB:CT, Jin vs. Litchi is a 5-5 matchup, Buppa (arguably the best Jin player) tended to struggle with it. Jin vs. Ragna was slightly in Jin's favor, but Kaqn (a well-known Ragna player) excelled at defeating Jin players. In SSF2T, T.Hawk vs. Ryu was vastly in Ryu's favor (usually listed at 80-20, Ryu's favor!), but Damdai could eat Ryu players for lunch.
With generally at least 8- and sometimes over 20- characters to choose from, and with fighting games being a genre that are distinctly inferior online due to lag, generally a player is only going to know the real ins-and-outs of facing the characters that their local opposition play; other matchups, they'll have a rough idea of what to do, but they won't be nearly as solid on it.
Regardless, I don't think ES is really a good idea for any game (if game results are so consistent that starting from a disadvantage is not crippling, than a plain BO3 should be reliable enough) -- simply that it's extra bad for SC. Maybe it's extra-bad for SC and FG equally. I definitely DO think that in a death match style FPS like quake that ES would play less of a role. This entire section of contrasting games was mostly me trying to figure out why MLG would think ES is a good idea in the face of constant complaints, and highlighting that different games require different tournament structures (for example, I'm passionate about the idea that SC tournaments should force diverse matchups early in a tournament -- the same logic could apply to FG, but be a bigger deal because all the characters constitute more diversity to cover).
I'm actually a fan of the extended series rule, simply because it means that you can't have situations where the tournament champion has a losing record vs. someone he faced in the tournament- for an example of this, see the King of the Beta tournament, where Idra had a losing record against Tester, yet still won the tournament because of an arbitrary "well, he won at the right time."
|
On May 25 2012 11:13 Obscura.304 wrote:
In Starcraft, we can take a step back and say "ok, at 6 minutes into the game, Terran can be doing these things...".
this is not the same thing, because at several points in the game I continue to learn more and more about what my opponent is doing and then whittle down his available options based on what i see. this isn't the same thing as learning his tendencies or style -- if there's no robotics bay, then i'm not going to see immortals. period. unless he's hidden the robotics bay!
You're wrong on that estimation. For instance, in BB:CT, Jin vs. Litchi is a 5-5 matchup, Buppa (arguably the best Jin player) tended to struggle with it. Jin vs. Ragna was slightly in Jin's favor, but Kaqn (a well-known Ragna player) excelled at defeating Jin players. In SSF2T, T.Hawk vs. Ryu was vastly in Ryu's favor (usually listed at 80-20, Ryu's favor!), but Damdai could eat Ryu players for lunch.
If this is something which is very common (like the examples you gave are not really weird outliers) then yeah, it creates a similar situation with FG and SC2, if not stronger in FG. You would mitigate this by ensuring players face a diverse set of opposition at some point and avoid the possibility that one player's bracket draw is super beneficial for his character or style (for example, Leenock playing mostly ZvT during his run from Code A to Code S finals).
I'm actually a fan of the extended series rule, simply because it means that you can't have situations where the tournament champion has a losing record vs. someone he faced in the tournament- for an example of this, see the King of the Beta tournament, where Idra had a losing record against Tester, yet still won the tournament because of an arbitrary "well, he won at the right time."
well it's not really arbitrary, both players knew the the value of each match. there's a certain amount of outside the game strategy that go into dealing with a multi stage tournament even if you won't face the same opponents -- maybe you have a really good zvp build you hold back on during the group stage and accept that your seeding will be worse but your performance in the playoffs will be better. on the other side, what happens in ES is that you face someone not having any idea if you'll play again and how important that game will be.
i'm not really bothered at the idea of one player having a losing record vs another over the course of a tournament. there are lots of tournaments ongoing and match results are going to vary widely along with how the match worth varies widely. just because they happened to lose inconsequential games doesn't bother me just like it doesn't bother me if they lose a series in one tournament this weekend and then win it next weekend. i'd rather that the players just know what's on the line exactly when each match goes on. but even without extended series, players can finish a tournament having a losing record vs someone if they simply lose and then never play again -- in this last gsl code S, mvp went 2-2 with ryung and 1-2 against HerO during group stages, for example.
thanks for all the examples!
|
 Curious how many more player handle puns you have in your sleeves.
|
Even though I'm  Keen on puns, I think I just got Lucky with this one.
|
Good read overall on your thoughts on ES. It can be annoying to see how ES plays into an MLG and how it affects the players. The real question though is what would be a better option at this point??
|
On May 25 2012 22:57 TheRealNanMan wrote:
Good read overall on your thoughts on ES. It can be annoying to see how ES plays into an MLG and how it affects the players. The real question though is what would be a better option at this point??
Thanks.
Part of the reason my first two articles are on tiebreakers and extended series is because I think we're better off just removing them, no replacement necessary. As I keep blogging on tournaments I'll include a lot more in the way of suggestions or alternatives (better to call them alternatives, since I don't believe there is "one perfect method" but it depends greatly on the context) since I'll be discussing more "gray areas"
|
Just heard if Sase makes it to finals, he starts down 0-2... Extended series crushing dreams 2012
|
|
|
|
|
|
EPT
