Post by awall on Jul 26, 2009 0:34:50 GMT -5
This thread is intended to stir up some discussion regarding the following question:
1) Is Paralysis broken? Not "cheap," not "annoying," not "unfun," but downright broken.
Resources and other related topics:
Your own custom spells - Taliesin outlines several problems with paralysis and suggests Feeblemind (FSD) as part of a replacement to it. Alternatives are discussed.
Enrage - A Practical Assessment - Exdeath outlines several issues with the game and notes that a great many of these could be fixed by changing/removing paralysis.
David Sirlin's website - A treasure trove of useful game design information. I wanted to link specifically to his article about when and why something should be banned, but after going to the site with the intent of pulling up a link to it, I got sidetracked and read other random articles for an hour, and finally just decided to quit procrastinating and address para already.
My answer to the question is: Yes. But I'm only 60% confident here.
A move in a game is obviously "broken" when its presence in a game reduces the game to only a single viable strategy: to use the broken move over and over. In this case, there really isn't a counter, so even if you know exactly what your opponent is going to do, you still get beaten by it. Examples of this would include Tolarian Academy from Magic (during the brief period it wasn't banned in Type 2) and roll cancels in Capcom vs. SNK 2 (never actually played it, but I'm led to believe this is the case from what I have read). Paralysis thankfully does not exhibit this behavior.
A move is more subtly "broken" when it cannot be used 100% of the time without fear of retribution, but it is still the optimal thing to do 90% (or some other suitibly large percentage) of the time. In this case, there exists a counter to the move, but the risk-reward ratio of the counter vs. not doing the counter is unacceptable; you can counter the broken move over and over, but screw up once and you're dead. As an example, consider what would happen if stabs did 16 damage. Is it a guaranteed win to do nothing but stab? No, there exists a counter (shield). But the risk-reward outcomes drastically favor the guy who throws out a lot of random stabs. Even if his opponent successfully shields the first 15 of them, all it takes is one bad guess and the stabber wins the game. Obviously stab is broken here.
Paralysis is not quite this blatant, but I believe it is just a bit too dominant. My reasoning is similar to the example with killer stabs above, though the case of para is far less severe.
Let's look at a concept that Sirlin refers to as "yomi." The basic idea is that there is one move that you really really want to do (M). I know you want to do it, so I can make a counter move (C1). If you predict my counter, you have a different move you could do that punishes me (C2), and if I predict C2, I have yet another move (C3) that beats it. If you think I'll do C3, you should just go ahead and do M, because we've already established that M is what you really want right now. (See the third link at the top of the page for a much fuller explanation and several well-written articles about yomi.)
In Warlocks, you see situations like this all the time. Consider PSD vs. xWP when the second player has an Ogre. Move "M" is for the first player to charm the Ogre. It's what he really wants. Knowing this, the second player casts counterspell on the Ogre (that's C1). But since everybody knows that the "right" choice is to protect the Ogre, the first player will generally charm his opponent (C2). If the second player is feeling really crafty, they can counter themselves (C3)... but if the first player predicts this, he'll steal the Ogre. There's a great deal of risk-reward going on here, and the situation favors the player with the Ogre, but either player stands to gain a great deal by taking risks and reading their opponent properly.
My problem with para is that, in my experience, these types of junctions favor the player with para far more often than is reasonable, given the ease of gesturing three F's in a row. When I play Succat, I am reasonably confident that he is going to open F/P or F/S and try to go for paralysis every chance he gets. This should give me a huge advantage against him, but in practice it doesn't. There are relatively few moves that really punish para very badly, and those that do are often very, very risky on the off chance that he doesn't go for para. That is, the risk-reward diagram for the situation where one player (A) has a para chain going often looks something like this:
A casts para, B does a "normal" spellflow = moderate advantage to A
A casts para, B predicts it and does a counter = small advantage to B
A doesn't cast para, B does a "normal" spellflow = small advantage to B
A doesn't cast para, B does a counter = moderate advantage to A
Savvy people will notice that the risk-reward diagram for the Charm vs. Ogre/Counter similarly favors the guy with the Ogre. However, an Ogre is quite difficult to get, whereas an F-chain is simple enough that it really oughtn't be flat out better than so many other options. Note that none of the other three-gesture spells in the game has nearly this nice a risk-reward ratio. The best one is Amnesia, and even that has plenty of counters, especially once you're past the beginning of the game.
Of course, one could also make the case that I'm oversimplifying this. It is very true, there are many times that going for para is a dumb idea and will get you mauled. There are also times when the risk-reward table is more even, and happens to favor the player without a para chain. There are some skilled players that choose other moves even wher para would have been viable, and some people who spam it and still get destroyed. But none of this satisfactorily demonstrates to me that para isn't broken. In a game with 43 spells, when one spell claims an advantage such a large percentage of the time, and is relatively safe even when it doesn't, I think something needs to change.
To summarize: I believe that game situations where one player has paralysis favor that player far more than is appropriate for how difficult paralysis is to cast, and therefore paralysis is broken and leads to degerate gameplay.
Note 1: I apologize if my thoughts were not entirely coherent, or if I didn't provide enough evidence. I tried to add in some concrete examples, but I had a difficult time doing so. My arguments against paralysis are largely based upon my gut feel after two years or so of experience with Warlocks, and paralysis is close enough to the line between "very powerful" and "too powerful" that for every situation I could hypothesisize where para is unfairly good, it would be easy to show a similar one where it isn't.
Note 2: My objections to para should not be construed as an objection against the play style of those who spam it. Casting paralysis frequently has shown to be an effective method of play, and I applaud Succat for focusing upon (what he felt to be) the best option available. However, I believe that it is in the best interest of the game as a whole to alter it such that paralysis is no longer as overpowering as it is now.
1) Is Paralysis broken? Not "cheap," not "annoying," not "unfun," but downright broken.
Resources and other related topics:
Your own custom spells - Taliesin outlines several problems with paralysis and suggests Feeblemind (FSD) as part of a replacement to it. Alternatives are discussed.
Enrage - A Practical Assessment - Exdeath outlines several issues with the game and notes that a great many of these could be fixed by changing/removing paralysis.
David Sirlin's website - A treasure trove of useful game design information. I wanted to link specifically to his article about when and why something should be banned, but after going to the site with the intent of pulling up a link to it, I got sidetracked and read other random articles for an hour, and finally just decided to quit procrastinating and address para already.
My answer to the question is: Yes. But I'm only 60% confident here.
A move in a game is obviously "broken" when its presence in a game reduces the game to only a single viable strategy: to use the broken move over and over. In this case, there really isn't a counter, so even if you know exactly what your opponent is going to do, you still get beaten by it. Examples of this would include Tolarian Academy from Magic (during the brief period it wasn't banned in Type 2) and roll cancels in Capcom vs. SNK 2 (never actually played it, but I'm led to believe this is the case from what I have read). Paralysis thankfully does not exhibit this behavior.
A move is more subtly "broken" when it cannot be used 100% of the time without fear of retribution, but it is still the optimal thing to do 90% (or some other suitibly large percentage) of the time. In this case, there exists a counter to the move, but the risk-reward ratio of the counter vs. not doing the counter is unacceptable; you can counter the broken move over and over, but screw up once and you're dead. As an example, consider what would happen if stabs did 16 damage. Is it a guaranteed win to do nothing but stab? No, there exists a counter (shield). But the risk-reward outcomes drastically favor the guy who throws out a lot of random stabs. Even if his opponent successfully shields the first 15 of them, all it takes is one bad guess and the stabber wins the game. Obviously stab is broken here.
Paralysis is not quite this blatant, but I believe it is just a bit too dominant. My reasoning is similar to the example with killer stabs above, though the case of para is far less severe.
Let's look at a concept that Sirlin refers to as "yomi." The basic idea is that there is one move that you really really want to do (M). I know you want to do it, so I can make a counter move (C1). If you predict my counter, you have a different move you could do that punishes me (C2), and if I predict C2, I have yet another move (C3) that beats it. If you think I'll do C3, you should just go ahead and do M, because we've already established that M is what you really want right now. (See the third link at the top of the page for a much fuller explanation and several well-written articles about yomi.)
In Warlocks, you see situations like this all the time. Consider PSD vs. xWP when the second player has an Ogre. Move "M" is for the first player to charm the Ogre. It's what he really wants. Knowing this, the second player casts counterspell on the Ogre (that's C1). But since everybody knows that the "right" choice is to protect the Ogre, the first player will generally charm his opponent (C2). If the second player is feeling really crafty, they can counter themselves (C3)... but if the first player predicts this, he'll steal the Ogre. There's a great deal of risk-reward going on here, and the situation favors the player with the Ogre, but either player stands to gain a great deal by taking risks and reading their opponent properly.
My problem with para is that, in my experience, these types of junctions favor the player with para far more often than is reasonable, given the ease of gesturing three F's in a row. When I play Succat, I am reasonably confident that he is going to open F/P or F/S and try to go for paralysis every chance he gets. This should give me a huge advantage against him, but in practice it doesn't. There are relatively few moves that really punish para very badly, and those that do are often very, very risky on the off chance that he doesn't go for para. That is, the risk-reward diagram for the situation where one player (A) has a para chain going often looks something like this:
A casts para, B does a "normal" spellflow = moderate advantage to A
A casts para, B predicts it and does a counter = small advantage to B
A doesn't cast para, B does a "normal" spellflow = small advantage to B
A doesn't cast para, B does a counter = moderate advantage to A
Savvy people will notice that the risk-reward diagram for the Charm vs. Ogre/Counter similarly favors the guy with the Ogre. However, an Ogre is quite difficult to get, whereas an F-chain is simple enough that it really oughtn't be flat out better than so many other options. Note that none of the other three-gesture spells in the game has nearly this nice a risk-reward ratio. The best one is Amnesia, and even that has plenty of counters, especially once you're past the beginning of the game.
Of course, one could also make the case that I'm oversimplifying this. It is very true, there are many times that going for para is a dumb idea and will get you mauled. There are also times when the risk-reward table is more even, and happens to favor the player without a para chain. There are some skilled players that choose other moves even wher para would have been viable, and some people who spam it and still get destroyed. But none of this satisfactorily demonstrates to me that para isn't broken. In a game with 43 spells, when one spell claims an advantage such a large percentage of the time, and is relatively safe even when it doesn't, I think something needs to change.
To summarize: I believe that game situations where one player has paralysis favor that player far more than is appropriate for how difficult paralysis is to cast, and therefore paralysis is broken and leads to degerate gameplay.
Note 1: I apologize if my thoughts were not entirely coherent, or if I didn't provide enough evidence. I tried to add in some concrete examples, but I had a difficult time doing so. My arguments against paralysis are largely based upon my gut feel after two years or so of experience with Warlocks, and paralysis is close enough to the line between "very powerful" and "too powerful" that for every situation I could hypothesisize where para is unfairly good, it would be easy to show a similar one where it isn't.
Note 2: My objections to para should not be construed as an objection against the play style of those who spam it. Casting paralysis frequently has shown to be an effective method of play, and I applaud Succat for focusing upon (what he felt to be) the best option available. However, I believe that it is in the best interest of the game as a whole to alter it such that paralysis is no longer as overpowering as it is now.