Execution in Fighting Games
Monday, July 16, 2012 at 6:08PM Here's an article about the role of execution in fighting games, by James Chen. I think the title and conclusion are kind of misleading because it does not really discuss the role of exeuction in fighting games, but rather the role of how different move commands affect the game.
Let's start with the good. Every example given is helpful to know, and I think each one is a correct example. In each case, Chen shows how the motion for doing a certain move being one thing rather than another thing affects gameplay and that this is usually for the better. Yes, a dragon punch motion and a reverse dragon punch motion do have different effects, and cause the moves to be used differently. Players should take that into account whether they are playing *as* such a character or *against* one. Good stuff here.
The problem is that that's not really what people mean when they talk about execution in fighting games. It's a very narrow, cherry picked kind of thing that doesn't fairly represent the topic. It also lead Chen to make this conclusion:
"Execution isn’t just about performing your combos. It’s also largely about knowing what your opponent CAN perform in given time frames. That ADDS to the mind games and the gameplay, not detracting from it."
I agree completely that the different motions chosen for moves gives another layer of things to think about and that that kind of variety is interesting relative to a game that had no such variety. The thing is, "execution" in general (not these very specific examples) has the opposite effect and it reduces strategy, relatively speaking. The more a game is about the difficulty of making your character do what you want to do, the necessarily less it is about strategy (that is, making good decisions).
This is why it's not a good idea to make special moves really hard to do. Make them take some *time* so some prediction is needed (even a few frames of prediction), yeah that's great. Make them start at a particular place on the joystick, such as a reverse dragon punch, and that affects how they're used, right. But to have some tiny input window to make them hard to do even when you have decided you want to do them, that's taking away emphasis on strategy. Making a game where the command to throw is secretly an option select tricky thing that you want to do basically always is another way to put more weight on dexterity that necessarily reduces strategy. Choosing commands that overlap too much (for example, ST Cammy's hooligan throw and spinning knuckle) puts more emphasis on dexterity than the decision of choosing the right move. So to increase strategy slightly, it's better to make those not overlap (and in HD Remix, they don't).
Making a game such that bread-and-butter combos require 1-frame linking is another great example of reducing the importance of strategy. In a recent stream, Chen himself said that in SF4, if you can't do Sakura's 1-frame link combo, you shouldn't be playing Sakura. I agree! That goes to show how strongly execution is favored over correct decision making / strategy in the case of SF4 Sakura.
We should really be striving to reduce execution requirements as much as possible while keeping the nature of the game intact. That is, making all dragon punches a single button press would reduce execution, but it would also actually ruin a bunch of strategy stuff by making them too reactive and not predictive enough, so we shouldn't do that. That's not a case where reducing execution helps, so I'm not talking about things like that. I am talking about sequences or moves that are hard apart from any strategic consideration. Like Sakura having 1-frame links as a critical thing, instead of being a character anyone could play. (You don't even need to change the power level of the character or reduce any strategy here, it's just a matter of being more inclusive as to how many players get to participate in that strategy.)
I know there's a lot of execution fetishism going around, and that's unfortunate for a genre that many would like to point to as a strategy genre that happens to have a dexterity requirement to play. Rasing the dexterity requirement above the minimum needed to make it all work just subtracts from the importance of strategy while excluding people. I'd like to see more love for an inclusive approach, as that restores more power to good decisions while inviting even more players to participate in those decisions.
Reader Comments (36)
I think the reason behind SF4's 1-frame links has nothing to do with anyone thinking it's enjoyable for players. It's there to let pro players do something obviously visible that the average Joes can't, so that you have a reason to go spectate them (and pay for the privilege), instead of just some watching some random guys. It's the most hamfisted "solution" possible, unfortunately. They made a similar compromise with ultras (worse for players, better for spectators), so I wouldn't be surprised if that was the driving force behind this decision as well.
"Execution" as something that gets between your decision-making and what actually happens in-game is too broad a term, I think. Here's a bunch of subtypes I see, and you may say "want" to some, "do not want" to others:
- Reaction times: "I want to SRK that guy in the face AS SOON as it's possible." Getting rid of this would mean creating a turn-based game, where you have all the time in the world to evaluate the current situation and make the proper decision in response.
- Distance perception: "I want to stand JUST outside of the other guy's poke range." This would be addressed by having distance measured in large, discrete steps, so there is no ambiguity in whether something's in range or not. In other words, locking players to a grid.
- Muscle memory: "I want to do the input EXACTLY so this beastly combo actually works." Discussed often enough, I'll not waste more words on it. Possibly except for "DO NOT WANT".
I think SC2 is in a fine location on the strategy/execution axis, actually. As in, yeah, this is as low as execution is going to get, if you want to preserve the game's nature, elegance, responsiveness, etc. Even as it is, they've made some choices to make execution easier, despite having a negative impact on the gameplay (smart pathfinding + large control groups = deathblobs *sadface* ). It's not low-execution still, but the APM required to do pretty much everything you want to do is within reasonable bounds now, at least (i.e. if you can type at a decent speed, you're good).
Sirlin, I think you consider improving on it simply "doable" because you haven't actually tried yet. :P Being more of a fighting game player, I would guess that you don't have an equally deep understanding of mechanics and design when it comes to RTSs, and so are less perceptive of the subtle, difficult to solve issues that would arise.
I wouldn't mind being proven wrong, of course, I would love to see an improved version of Starcraft. So if you think you can do better, I'd be interested in seeing what you come up with to do so. (After getting Fantasy Strike out the door, obviously.)
The real question is how do you make stuff like kara-cancelling not work. I look at Skullgirls and its IPS, and folks figured out how to game that. One idea I had was for the game to hard remember the position when you started to press a button, and make animation for moves start from there if it's a move that moves the back of the hitbox forward or something along those lines.
The GG/BB style proration is the best method- the first move should really determine the damage scaling for the most part- reward folks for landing heavier CH's.
Specifically what questions do you have? Maybe the community has some ideas or theories on how to prevent the problems you're seeing.
BTW, Have you considered kickstarting your fighter project?
As for execution, I think the big thing is casual players need to feel like they can do the movelist, and intermediate players need to feel like they can do the vital combos. If an advanced wombo combo does 5% more damage than the intermediate version, that's not as bad. Also, there needs to be no execution technique that dumbs down the game at high level (especially defensive techniques). My big execution complaint with SF4 is how some of the option selects do this. VF does it right for the most part, as the big defensive techniques can be blown up hard if predicted.
Starcraft is the game of economy. You have to manage your resources well and one of the most important ones is your concentration - even for pros complicated micro takes a lot of concentration, and sometimes its not worth it. Especially on lower level if your army is doing reasonably fine on its own its more important to spend a few APMs on managing your base then ten times more on micro that will improve their efficiency only a bit. Without dexteriety component it would be a completely different game. Having said that i do agree that Starcraft II does have huge reflex/ muscle memory component by design, it was expected having Brood War as a predecessor. Even if not absurd APM new players are obsessed with, you do have to learn your build by hard and learn to execute it as fluently as possible. (not pure APM, but requires a ton of training) My buddy beat me (and other players online) using touch pad and having relatively low APM - he just had his build tailored to not wasting APMs practiced and it was good enough. I got to platinum after playing a few months. I do not claim to be good SCII player, but i'm not a complete newbie either.
I'm not that sure having everything doable 100% is the best way. If you want pure strategy no reflex you can play a boardgame (and i love boardgames, there should be more boardgame-like electronic games, so many great mechanics out there in boargaming world). And if you love pure elegance and strategy just play Go - great game.
The real problem is how to design a game so that beginner who doesn't wont to spend so much time doesn't get completely deprived of games depth. If the number of strategies that make sense for him is too small playing the game will be awful experience of repetitively doing one thing - and that doesn't sound fun. I don't think its necessarily has to be even field for ppl who dont have what game requires... Some games are games of chess-and-cooking and its fine with me. But if you need to cook decently otherwise you cant even make a simple move this game stops being fun for too many ppl who wanted to enjoy it.
I really start to wonder how would strategy focused fighting game feel like. Seems hard to design as even if moves themselves are really easy you still have timing of reaction to them and interaction of opposing moves to time. A tool to prototype fighting games would be nice here for some proof of concept, i wonder how hard is it to do a demo.
I think both have their place. I prefer games with lower execution amounts because I haven't got time to practice them. I also like watching esports and I found that MvC3, SC2, and Dota2 are among the most exciting for me to watch. I think a lot of my interest has to do with the pro players being able to do things that I cannot.
To contrast, I personally very much enjoy playing puzzle strike, but it is my impression that high level play consists of counter-picking characters/chips and requires considerable meta game knowledge. In execution style games you tend do see a lot more strategies that a player can force. As in, it shouldn't work, but this player makes it work because he's just that good.
Wow pkt, I disagree really strongly that a starcraft game "needs" to have the kind of APM requirements of the actual starcraft to have the strategy it has. I actually can understand how you or anyone could even think that. Blizzard intentionally chose a high APM game, and put in stuff to specifically reward it. I think it doesn't take much imagination to think about not doing that. But you revealed your defeatist attitude about it already by saying selecting more than 12 units leads to worse strategy. It really doesn't, it leads to a closer link between what you want to do and actually being able to do it. The "deathball" problem is like some totally other thing that could be solved without making intentionally bad UI to select units. I think they are working on solving that right now, even. I have zero interest in working on this APM in RTS games thing, but to say you *can't* make a lower APM game with equal or MORE strategy is actually baffling to me and very bold claim. It's certainly possible, and it was clearly not Blizzard's goal. Just think if it had been.
Alstein, sure but fyi it's "just putting on kickstarter." You have to have a plan first about how to get all the way to shipping. Exactly how much will it cost and what's involved. It's a multi-million dollar thing to make fighting game these days, and so there'd have to be planning about who is the producer, who is the art team, what's the schedule, etc. Huge project to put something that big on kickstarter, and I just don't have the pieces or even time to do that right now, with Yomi and another card game in the works at the same time. So that's why we'll continue just slowly improving what we have on the fighting game with whoever shows up to help, for now. Reaching a demo-able state would sure make going to the next step much easier anyway.
zero: there's a much easier solution to the combo thing you said than an elaborate air-control system like smash. Just make combos really short like in SF2. They are over in like 1 second. This is only a problem to begin with in games like MvC3 or something where you are stuck in a non-interactive state so long that it can far outnumber the number of seconds you are in control of your character. Also, put in a burst mechanic of some kind and the person getting combod has to make a decision. I'm not really worried about this aspect. (I'm worried about finding awesome 3D artists, lol.)
I think StarCraft 2 has a lot of artificial things that push up the execution barrier.
Here's the most obvious APM drain: It is optimal to always go to your production building right as a worker/unit/research finishes and you start a new one right when it finishes. This is typically 2-3 actions per event, and in the midgame can easily exceed 10% of your APM.
A simple solution: you can queue up workers/units/research(!) in their respective buildings. They don't consume resources until they actually start being produced. If you want to continuously produce workers or units, you can right-click to enable auto-production which keeps the queue empty but automatically starts producing the relevant unit as soon as the building is not producing anything else and you have the relevant resources (so you can still interject other units in the queue).
This removes a huge portion of the difficulty in producing your workers and units, but at the very, very highest level it has little effect (since you would generally not enable auto-production, since you might forget to turn it off and you don't want to run out of resources by producing a marine you don't really need a few moments before producing an SCV you DO really need).
"LG: if that example combo weren't so hard, then everyone could participate in the gameplay where that combo can be used."
I don't get that. Are you asking that person A who isn't putting in as much time and effort into the game as person B should be able to beat B on a regular basis? If person A invested that time, then he would be able to participate in that gameplay. Strategy is something you have to (partially) train in p-mode, why shouldn't execution also be a way of testing someone's skill?
I always thought execution is part of fighting games, and if someone doesn't like it then they'll have to play another game. I'm not trying to change the ruleset of chess to "you can move any piece you want any amount of times, as long as you can physically do so in 3 seconds" just because I think it should be less about strategy.
LG: It does not follow that you should get an advantage in a game "because you put in more time." You should get an advantage for being BETTER at the game, yes. If you can reach skill X in 10 hours and I can reach it in 100 hours, then so be it, but it's the skill, not the time, that we care about.
This brings us to the question of which skills we care about. We agree that we don't care about the cake baking skill in Chess. So let's throw that out, it's stupid to test. Ok, what about your combo skill in a fighting game? If this is to be a strategic game, then yeah I don't care about that skill. It has nothing to do with strategy, and rewarding it only demphasizes the strategy of a game. The more a fighting game reward combos, the more it is "dumbed down," in the strategic sense. You are certainly free to like this kind of dumbing down, but I think it's just more interesting if more emphasis is on strategy.
You also said that if people don't like it, they can play a different kind of game. I've addressed that already. A certain amount of dexterity is needed to make the genre work, so I am not saying to remove that. I'm saying if you already have it working, and then you add a whole bunch MORE execution test on top, you're just making a worse strategy game that tests this other skill. Some people like that other skill, ok fine. I'd rather not focus on it and not add a bunch of tests for it.
The thing about combos is that they are fun, assuming the game handles well. (I find combos fun to execute in marvel 3, and really painful and clunky in SFxT for all non-Ryu characters.)
There is also the strategic element of meter management and the RPS Team Ariel Combo buried in MvC3's mechanics. I'd like to see a game that manages to maintain these elements (meter management, pressing your luck to extend your combo) without having the long sandwich combos that MvC3 has normally.
Kdansky: "Another issue nobody comments on: Combos are boring. There is no game happening, it's just waiting until we can continue. Imagine Basketball, but after every score, they do a little dance of a minute or two like in soccer. It would be the worst game ever."
zero341: "The 'problem' with comboing with people who don't enjoy raw executition is that from the beginning to the end of a combo, there's no actual interaction between the players (it doesn't matter what the player being comboed does)."
BlazBlue (and some other ASW games) has the Burst system to allow you to break your opponent's ongoing combo, so unless you are out of Burst points there is still something you can do while being comboed. Deciding when to Burst can become an important part of your strategy.
It's so unfortunate that Arc System Works already solved so many of the problems that people are complaining about in fighting games (including the notable non-gameplay problem of badly designed button config UI that slows down tournaments significantly), yet their games still remain relatively obscure outside Japan...
Yeah BT, those things are great. Choosing whether to use meter or not, or choosing to end in a good situation for you or instead end with something else that does more damage. Having at least some combos gives more importance to positioning, which is cool. For example, a jump roundhouse at the very tip that barely hits and one that is really solid and close give a different reward: the first one can't combo into more stuff but the second one can. So that gives you a lot of reason to try to get into that right position.
There's plenty of positive aspects to combos there, though none of the things listed above require a layer of hard stuff added. You can get all that without 1-frame links inside bread-and-butter combos, or making everyone kara-throw instead of regular throws with more range to begin with, etc.
Mostly talking RTS, but maybe this applies elsewhere: If you want more strategy and less execution, why isn't slowing the game down the correct answer?
Andy: that's a good question to ask, but I don't think it really is the answer. Easier to answer in fighting games, that making the game slower lets people react to even more and the relative importance of prediction goes down. We experienced this in Super Street Fighter 2, which was slow and not really liked. Even apart from the lower fun factor it had, it just makes jump ins too good relative to fireballs, as that whole thing is tuned based on what human reaction speed tends to be, so slowing it down kind of messes things up.
As for RTS, maybe not a great answer either. I mean there could be something to that, but it seems the problem lies more in the difference in reward of very high APM compared to very very high. Even on a slow speed, I would feel like more and more is still always better to do. Some sort of cap is possibly a better place to look. Like a thought experiment (not necessarily a real suggestion) is to imagine you had a click resource of 60 per minute or something. You really have to think about whether you want to spend all those in one place on the map, or spread it around. There could be some other less artificial way of getting at that same idea than a measurement of clicks, so that's more of a brainstorm thing to spark other ideas.
I'd like to point out that my claim wasn't that lowering execution would make it impossible to retain just the strategy in Starcraft, but also the game's nature, elegance, responsiveness, etc. The solutions you proposed (YEARS ago, so possibly not relevant anymore) certainly seemed to have these problems, and I wouldn't say you're lacking in imagination. If you want to understand how I could think this, it'd be easier to talk specifics instead of generalities. Maybe we'll find out I don't actually think this, heh.
But you revealed your defeatist attitude about it already by saying selecting more than 12 units leads to worse strategy.
The point isn't that the latter is unavoidable, but that it's an unintended consequence. I'm saying your obvious solutions to reduce execution can similarly lead to new problems, which will require non-obivous solutions. I don't think it's defeatist to say that this is actually going to be a rather difficult problem, when you can't even smarten up the pathfinding without it messing with the gameplay.
I have zero interest in working on this APM in RTS games thing, but to say you *can't* make a lower APM game with equal or MORE strategy is actually baffling to me and very bold claim.
Less bold a claim than saying you can do it BETTER than Blizzard, who have been at this for literal decades. :P Then again, you say Blizzard aren't even interested in solving this problem (I disagree, they did put in automine, MBS, smartcast, etc.), so this isn't going to be very convincing for you.
As for the problem being mostly how the game rewards going from "very high APM" to "very very high APM". I'm not sure, maybe? But if you can do pretty much everything you want with "reasonably low APM" anyway, isn't worrying about that stuff optimizing for the 1% of instead of the 99% ?
@ajfirecracker: Thing is, clicks are also a resource. Auto-queuing would mean making things cost 0 of that resource. It's not like incompressible clicks such as this only add execution, either. Here's an article dealing with the subject: http://www.sirlin.net/blog/2009/4/13/uc-berkeley-starcraft-class-week-10.html
Sirlin said: "For example, a jump roundhouse at the very tip that barely hits and one that is really solid and close give a different reward: the first one can't combo into more stuff but the second one can. So that gives you a lot of reason to try to get into that right position."
This annoys the hell out of me. It feels like a really annoying barrier to execution. I know exactly what the combos doing, I'm hitting the right buttons and all I get is nothing. I can see the tips not leading into the combo, but in HD Remix it feels really strict and unreliable. Something like a solid round house jump kick over the head should work into a roundhouse sweep but it feels more like I have to hit chest level for it to combo. The visuals don't feel like it makes natural sense as to why the combos don't go through sometimes.
I still can not land the 3rd hit with Vega with : Jumping Hard Punch, Crouching Medium Kick, Link into Crouching Medium Punch. Never get the medium punch in there. Why's it have to be so tight. I can do special moves to a pretty reliable degree, things like the Vega combo I have so much trouble with in HD Remix.
Blazblue takes that combo execution and makes it a lot more reliable but at the same time they have to make combos too damn big, and then that becomes unreliable to remember. I'm so tired of not playing the game a few weeks and I can no longer remember the extended 17 hit or so Ragna combo. And real good Ragna players seem to run around 25 to 30 hits.
I wouldn't complain so much about execution if there truly were games that had what feels like true low level execution.(Especially a 2D one) Maybe some of VF 5's characters are reliable low execution but it really feels like most fighting games at the forefront suffer from execution issues.
Spurn: L2P I guess. My example is as simple as you can get. Jump roundhouse when close lets you do a second move. If you're too far, the second move won't reach in time. If your complaint is the hitstun being too short in some particular game, that's not a comment about the nature of my example, it's a comment abou the hitstun in that particular game. The example is a very good one, it illustrates why the reward for landing an attack differs by distance, and is about as simple as you can get.