I spent most of the weekend creating tiny gameplay demos, trying to figure out what was fun and what wasn't about the much-vaunted "fighting game".
I'm not talking about how to balance them. I'm talking about the core idea of how they play. What makes them fun?
In the end, I've come up with a new theory. I think that fighting games are like Mahjong. That is, they are a matter of weighing risk and reward in a complex environment.
We talk a lot about how fighting games are about timing or distance control or predicting the enemy or whatever. But, really, all of those things are just a way to decide what the risk and reward levels are, and how to tweak them.
You stay hovering just out of the enemy's short, fast attack range because it means the enemy would have to do a heavy attack to hit you. Heavy attacks are inherently riskier, so you're basically saying "oh, you can attack, but if you do, I won't have any problem blocking it or stepping out the way and slamming you with a counter."
Conversely, if you stand at long range and spam fireballs, you're accepting a much lower reward in exchange for much lower risk. And if you use super juice to do a special attack, you're either lowering the risk (by adding block-breaking and range) or upping the reward (by dealing a lot more damage).
All of the mechanics of a fighting game are there as avenues towards risk and reward. The player juggles these. The player decides how to raise the enemy's risk, lower their own risk, raise their reward, lower the enemy's reward... and there are a lot of possible approaches.
Thought of like this, the omnipresent super meter which I always hated becomes an obvious part of the game. The super bar's myriad uses always grated. "Pay 0.3 notches for a cancel! Do a half-super or a full-super or a level 8.1 super..." But from the perspective of risk wrangling, the super bar's use for things like combo-breaking and recovery-canceling are actually the primary purpose of the super bar. The flashy techniques are useful, but they're rather like a hammer - they give you the same tools but more oomph. Something like a recovery cancel or a combo break is a powerful new tool... if you can figure out how to use it.
This is made very clear in the Naruto fighting games, where teleportation dodges are the name of the game. Much of the melee combat in the game revolves around weighing whether you can afford to use your teleport dodge, or whether you can force the enemy to use his, or whether you can trade 1 for 1 until you get the final attack chain in... For example, if you have one teleport dodge remaining and the enemy has four, you can't really go in expecting to get a nice melee chain, because he'll teleport-dodge and land a chain on you. You could teleport-dodge again, but only the once... so at that point, rather than continuing to melee, it may be best to retreat or use a single-strike attack such as a super, which is much harder to teleport-dodge.
On the other hand, in DOA countering is a major concern. So a lot of matches feature waffling around at middle ranges to try and lure the enemy into striking with an expected, easy-to-counter attack. And then a lot of the game evolves into trying to avoid getting locked into a waffle combat, and maybe using dash attacks or feints or grabs - it's all a matter of watching the deny you the low-risk options and choosing to avoid the high-risk options my taking some mid-risk options. As a game with no super meter, the combat feels very pure, rather than having the technical weight of Street Fighter. The risk analysis is sharp and obvious.
You have all these tools to manage risk and reward, yours and the enemy's. In many cases, there's even several tiers of game - tag team games frequently allow the off-screen members to regenerate, so you may want to push the on-screen enemy to swap out and truncate the health of the off-screen enemy... the reward for forcing a tag switch is very good for you. From the enemy's perspective, they are being rewarded for simply not switching out. This is also true in Naruto, which has no actual tag-team members (just helpers), but still has that marching time component in teleport dodge regeneration. If you're out of dodges, you'll often stay at long range and play a stalling game with dodges and ranged attacks until you regenerate.
So... all the tools used to manage risk and reward. Let's break it down a little more precisely into categories of tools, risks, and rewards.
An ongoing reward is something like regenerating teleport dodges, off-screen people healing, or the slow march of a timer when you have the advantage. In these cases, you'll generally aim for stalling - often maintaining a long range, turtling, hiding, and so on. The enemy will generally feel pressured to use riskier attacks to interrupt your ongoing reward, so this is a very strong pressuring mechanic.
A solo reward is one which does not involve interacting with the enemy. For example, charging your super-bar by standing still. This is related to ongoing rewards in that it is often something the enemy feels compelled to interrupt, but this is different in that the reward is triggered by your actions. Therefore, you cannot properly fight while obtaining a solo reward. Because of this, solo rewards are often "stolen" in moments when the enemy is stalling or out of position. It's also not uncommon for both players to charge their solo rewards in a kind of mutual situation - this is especially common when the opponent doesn't think he can really interrupt you very well, or thinks he'd do loads better if he just had a little more juice.
An attack is a combination of risk and reward. Rather then considering each punch individually, attacks usually include whatever chains you can use, which can radically increase the reward without increasing the risk much unless there are mid-combo counters. Attack risk depends on a lot of factors: lead-in time, recovery time, priority, block-breaking, sneakiness of strike zone, predictability. Reward is almost universally how much damage the attack chain will do if it hits, plus factoring in kncokdown, stun, and wall hits as necessary.
Most attacks raise some kind of super bar, which is a category of reward. If you get hit, it also frequently raises your super bar. These are simply elements of the reward, put in to reward aggressive play and put in a touch of negative feedback. Instead of how much super they generate, you might break attacks into two categories: immediate and asynch.
An immediate attack is one where the avatar is constantly in the attack as long as it lasts. This is what most battles rely on heavily, and when considering which immediate attacks to make you need to consider priority, range, recovery, whether the enemy can interrupt it, whether it's too predictable and they're expecting it... Most immediate attacks are melee, but sometimes you'll run into some kind of beam spam attack where the enemy just stands around acting busy for the duration (Mortal Kombat had many of these).
An asynch attack is one where your avatar finishes acting (or never acted) but the attack continues. Street Fighter fireballs are a major example of this, as are assists in most tag team games. Synch attacks allow you to lay down complex lanes of fire, and are often very low-risk. Asynch attacks are a frequent tool of turtles and stallers, who like making a low-risk attack and also being free to react to however you react, lowering the risk even further. However, they can also be used to extend or punctuate normal attack combos, so everyone will end up using them to some extent.
A condition is a situation in the ring which shifts the risk/reward calculations. An ongoing reward is probably a condition, but this is more about things like pits, walls you might get backed into or blasted off of, water that slows you down, fire jets which pop up once in a while, and so on.
A super action typically has a very good risk/reward profile, having a high priority, high damage level, and forgiving range. Their downside is that they use up some kind of limited resource or require a difficult activation. In the old days, the techniques of the fighters were often considered supers: roll to forward, down, forward, all the kicks? Hold back for two seconds, then forward and punch? These are hard to do. They take time and effort. There's the risk you'll fail to trigger them, and there's also the risk that the enemy will attack while you're farting around.
These days we've transitioned mostly over into supers that use up an energy bar, making it much more tactical and much less about whether you have a triple-jointed thumb. Street Fighter's introduction of "gems" is not one I agree with, but shows another path to supers: make them require you to perform specific in-fight events, such as hitting an enemy 8 times or whatever. Countering could even be considered a form of super, as it is generally quite hard to pull off and gives you a tremendous edge.
As you might guess, while most supers are simply one-off moves that are very good, other times they are part of an ongoing risk/reward pattern. For example, recovery canceling can allow you to chain your combo straight into another combo. "Berserk mode" makes you faster and more dangerous, altering your risk/reward profile for all moves for a few seconds. Countering frequently leads into a volley of your own moves, rather than being useful just on its own.
There may be a lot of overlap between "super" and "attack", but, hey, that's life.
Anyway, I'm not saying this is the be-all end-all of fighting game design. But it does give me some ideas as to new prototypes I could build!