Lots of people have asked me why the Ultras in Street Fighter 4 work they way they do. I have no inside information on this, nor have I even seen what the designers might have said, but I have a pretty good guess. After walking you through that, we can then ask if we would include such a mechanic in a new, ideal fighting game. I actually don't know the answer, but I can explain the issues.
When I first saw the ultra system a long time ago, I though it was a terrible, terrible idea. My reason was that the last thing casual players wanted was a SECOND super meter. I mean this isn't Guilty Gear, it's supposed to be simple and elegant, and a new super meter in addition to the old one (which has four divisions and multiple uses) is going to be a confusing mess. I think it turns out there was more to the issue than I thought back then, so let's look at all the issues together and see if the overall effect is good or bad.
But first, let's explore how I imagine ultras came to be what they are in this cartoonized, fictional line of thought:
"How can we get casual players interested in this game? There's lots of ways of course, but one way would be really flash super attacks. Yes, that's a natural answer because we already have had super attacks in lots of other Street Fighter games, and now that we can use 3D animation, camera movement, and effects, we'll be able to make these look even better than ever. So far, this sounds great."
"But wait, how often do people really land super moves? Maybe not quite enough. Think of the spectators, they'd probably like to see these flashy supers a lot. We want to make sure that even beginners see these cool supers, so they can't be some rare trickshot that's impractical in a real match. One problem is that we have this idea about multiple uses for your super meter. You can use it to do a super move, to do ex moves (powered up versions of special moves), or to do ex cancels (like roman cancels in guilty gear, where you do a special move, then cancel it instantly so you can do another move in a combo). Anyway, there is some strategy in managing your meter, and we like that, but it also means that supers will be LESS frequent. Some players will use ex specials constantly, for example, and while that might be a valid strategic choice, it's ruining our initiative to show off the cool supers."
"How about...a SECOND super meter? The main super meter works how we said above, with three different uses. But this new meter could only involve supers...a new kind of super that we'll call ultra. The ultras will be the visual showcase of the game, with camera movements and elaborate canned animations between attacker and victim. We need to make sure everyone gets to use these ultras though. If your opponent overwhelms you, you don't even get a chance to build up the standard kind of super meter. But what if this ultra meter filled up when you GET HIT? Everyone gets hit!"
"In tuning this idea for the ultra, there was somewhat of a conflict. On the one hand, we really want you to fill up that ultra meter fairly fast. If we only gave you the ultra when you have 1% life, for example, it would be too rare to see it happen. But if we give it to you when you still have 50% life, it's just way too powerful of a tool to have. So instead of just an on/off thing, where you either have/don't have the ultra...let's give it to people at 50% life, but keep the damage low. As they get hit more, their ultra meter keeps filling up so that when they do get really low, like 5% life, then the ultra does tons of damage...maybe 50%!"
"This is also gives everyone a comeback mechanism, so even if you're losing, you're still in the game."
"Finally, we want the ultras to be kind of special in that you don't just short kick, short kick, ultra all the time. So we shouldn't let you cancel moves into them. They'll be stand-alone attacks. But hmm...it sure is fun to combo into them anyway, so how about you can juggle into them, throw into them, and things like that to make sure they frequent enough."
That's the line of thinking I imagine lead to the ultra system. I'm not trying to be positive or negative about it with that description, just trying to explain why it probably is how it is. So what do I actually think about it? One factor is my original objection that a second super meter is inelegant and something that exactly no one asked for. I thought Casual Joe would scoff at it or be confused. What I greatly underestimated is how good these ultras would look.
The production values on the ultras are incredible, and in my opinion are the best looking things in the entire game. The animations are great, solid, powerful-feeling. The effects are great. Some of the game's animation makes me cringe and Blanka's entire character model makes me want to cry, but these ultras really are amazingly good looking across the board. While I predicted Casual Joe would not be down with even more super meters (each with different mechanics...) it seems that Casual Joe's actual reaction is "wow these look incredible!" That also spills over into "this game looks incredible" and gives it an aura that makes Casual Joe completely willing to overlook lots of problems, even art problems, that might stick out.
I'll specifically point out Gouken's ultra as looking and feeling terrific. The shin shoryuken has always been a good concept for a move that feels powerful, but Gouken's version...with 3D camera movement, huge hit pause on each hit (power!), and great animation...made the concept of the shin shoryuken come alive in a way we've never seen until now. A+ on that, if you ask me. Incidentally, he can combo it every time (easily) off a throw, so you're sure to see it often.
So if we ask "did ultras accomplish their mission?" (or at least the mission I'm guessing existed on this project), I have to say the answer is yes. They make the game much more exciting to spectators, they draw in casual players, and they contribute to that aura of good graphics that helps people overlook other graphical problems. But what about the effect on gameplay?
The most obvious effect on gameplay is that ultras offer a comeback factor. I still can't figure out why the concept of slippery slope is so hard for some people to understand (they intentionally try to misunderstand it, I think). I wrote an article about that here. The short version is that most fighting games are slippery slope neutral in that the scoring (your health meter) is not related to your ability to score. I mean this in general terms, so ignore edge cases like block damage (if you have only a sliver of life left, you can no longer do the move "block", yes I know that.) But whether you have 100% life or 5% life in most fighting games, your moveset is the same. You still have jump roundhouse and throw, and so on. There would be slippery slope if the more you got hit, the more disabled your character became, as in Bushido Blade.
Street Fighter 4 adds the opposite concept, which I called perpetual comeback in that article. It's really called negative feedback, but that term sounds..."negative" (it's not, it's just a counterpoint to positive feedback) and also people mix up those terms all the time. Anyway, SF4 adds a comeback factor. There are some who are against this in general, and who say a "real" game punishes your mistakes (and therefore a "real" game has slippery slope!). This is where I point out that your mistakes are punished in a game with neutral slope because a mistake means your score goes down (your health bar). It's just that your score going down doesn't make it less likely for you to win by reducing your ability to attack, which would lead to first hits being way too important.
So neutral slope is good, I think, but what about this comeback force? Can it also be good? This is a very delicate question because it depends so much on specifics. I think a game having a mild comeback force is usually fine, though when it becomes too strong, it throws things way out of whack. Another thing to consider is how easily you can make a comeback anyway, naturally, without extra forces (like ultra combos) stepping in. Here are some rules of thumb: if defense is really strong in a game, comebacks are hard. If offense is really strong in a game, comebacks are more possible.
Imagine a game where offense is, for the most part, really good. In this game, throws have instant startup. Even if the opponent breaks a throw, they still take damage. If you just sit there (trying to sit on a lead), you are open to throws. Also in this game, knockdowns are powerful. You can't vary the time you get up, so that makes crossups more powerful. It's also hard to reversal attack (a bad way to design things, but just go with it for now), so attacking a rising opponent is good. Stages are fairly small so you can't run away. Finally, hit point totals in general are actually really low, so even when you're behind, you aren't THAT behind. You can always go on the offense because, well, offense is strong.
Now imagine a different game where it's much harder to naturally make a comeback. Hit point totals are higher in general so that means when you're behind, you're actually further behind than in the first example game. Throws are weak with 3 frame startup (or worse, 5 frame startup for Ochio throw). Throws can be escaped for zero damage. The playfield is really big, so running away is more possible. Knockdowns are weaker (meaning offense is weaker) because reversal attacks are incredibly easy and varying your getup time helps a bit to weaken crossups. (Well sort of...). Anyway, all those things point to the power of offense lowering and the power of defense increasing, relative to the first example game. This second game (SF4) needs some comeback mechanism, that's for sure.
Ultras do provide that. I think ultras have an unintended negative effect though: the ends of rounds were supposed to become more exciting, but instead they sometimes become more boring. One of my friends who played me the other day remarked that it was frustrating to him that some rounds he would beat me down, but then toward the end, he had to play very carefully and conservatively. He just gave me a 50% damage ultra that I could randomly throw out as desperation at who-knows-what time, so he had to switch playstyles to a more boring one, basically.
Another situation that occurred in a match against him was when I had him at just under 50% life. I was Rose and just got my ultra. I thought about how if I use it soon, I could hit him (which is easy, it's like perfect anti air and beats just about everything), but I would not kill him. So maybe I will hit him a few more times, then do the ultra for the win. But then I thought, wait, what if he hits ME a few more times instead? That will actually power up my ultra enough that it will kill him outright! While it sounds better to hit him than to get hit on purpose, it did feel like the next few hits hardly mattered. (Epilogue: I got hit a few times, I ultra'd, I won.)
So I've been wondering, if we were to make a new fighting game, the ideal one, would we include something like ultra combos? It seems the market has spoken on the issue of complexity of two meters versus awesomeness of these supers. Awesomeness won. Now imagine all the extra hype, press, casual interest, and spectator interest our theoretical new fighting game would get from ultras....(at lot, right?) and compare it to the extra sales and interest we'd get from making sure the ends of rounds between experts aren't boring because they don't need to shift to cautious mode as much. Even if we did improve gameplay by avoiding cautious ends-of-rounds, it doesn't sound like a very good bullet point on the box, does it?
And even more to the point, it seems that when looking at this purely from a gameplay angle, it's a better solution to make comebacks more possible in a natural way by making offense good. Games with good offense have proven to be fun over the years, while games that allow defense to win are boring to play and watch. (And note that even games with lots of offense like GGXX and ST have defense too, it's just that attacking is good.)
So from a gameplay standpoint on our new theoretical fighting game, we might be better of with no ultra system at all, but better offense (and fewer hit points, perhaps). Even if we had a game that allowed comebacks naturally...would we STILL want the ultras anyway? Again, the idea that you get hit to fill up a special kind of super meter means that you will always be able to do that super before dying. It's a way to ensure that spectators and casual players get excited at awesome super animations that happen all the time, so is it really worth it to lose out on those benefits even *if* you believed that gameplay suffers? That's the scary thought I wrestle with. I'm not against comebacks, but when the mechanism to facilitate them is extreme enough to introduce some ill effects, and when it's possible to allow comebacks anyway with overall shifts in the design, I just don't know. It's a collision between trying to make the best gameplay possible and trying to make a game popular and get noticed. Maybe a good solution (for the theoretical game, not for SF4) would be to have one super meter and that supers are amazing looking, but we simply accept that they don't happen quite as often as in SF4. I leave it as an open question.