Rule of Cool vs Frameworks

Executive summary

DECIDING WHAT TO BUILD IS VERY IMPORTANT

WE CAN JUST GO WITH VIBES

THIS IS CALLED RULE OF COOL

IT CAN WORK, BUT OFTEN IT DOESN’T

WHEN IT DOESN’T WORK, YOU HAVE NO TOOLS TO FIND OUT WHY

INSTEAD YOU SHOULD FIND PATTERNS AND THEORIES THAT DO WORK

WE CALL THESE FRAMEWORKS IF WE’RE FEELING FANCY

ODDS ARE YOU ARE DOING THIS AT LEAST SUBCONSCIOUSLY

IT IS GOOD TO MAKE THIS EXPLICIT SO YOUR FRAMEWORKS SPREAD AND IMPROVE

Vibes-based game design

How do video game designers decide what to put into their games? Well, they think of something cool, try and see if it works, and if it does, great! Moving on to the next cool thing.

You may think I’m joking, but that’s how a lot of games in a lot of places are made. And it can work very well! Many billion dollar franchises were designed that way. Usually there are implicit frameworks (more on those later) at work invisibly, hidden away in the informal knowledge of the group of designers, but they tend to be rudimentary guardrails (as in, this kind of thing isn’t good in our game) rather than systems of thought that could suggest further design space to be explored.

Rule of cool can absolutely work

A lot of the work of the group of people that made the Call of Duty games up to the original Modern Warfare II and who then went on to make Titanfall 1 & 2 and Apex under the Respawn moniker is “rule of cool” work. This was sometimes explicitly called out (although I can only remember hearing the phrase in a positive context from a producer now), and evidently it worked really well for them. This started to change somewhat in Apex’s later seasons and I don’t think I was alone responsible for this. There were a lot of other strong designers who saw the value in going beyond rule of cool.

But when it works, it really does work. More academic design can struggle to make the thing that is good for theoretical reasons appealing and cool. When you start from “wouldn’t it be cool”, you’ve got that part covered. Theoretically.

So what’s the problem?

Designing by rule of cool is like rolling the dice each time you put something into your game. Will it work and make the game better? Possibly! Will there be problems down the line? Who knows! Will the new thing have longevity, will it eat into the play rate of existing mechanics in negative ways, will it be more powerful in games full of novice players or in games full of expert players? We don’t know! And worst of all, one of the most common failure modes of rule of cool is “oh yeah that’s cool but we can’t really do that in our game”. I call that failing to live up to your promises, and it’s often worse than not making the promise in the first place.

A subtle distinction

There’s a pretty subtle distinction I need to make here. There’s a difference between “design by rule of cool” and “using something that seems cool as the starting point for an idea”. The latter is almost universally how we come up with new stuff. The important distinction is that there’s theories and frameworks and classifications of existing content to guide you and to suggest possible answers to the questions we posed last paragraph.

So what do we do instead

Instead of starting in the abstract, let me begin with an example. When I was on the team that designed the League of Legends character Taliyah, on day 1 of that process we had a small meeting. If memory serves, four people were in that meeting: a producer, a designer, a narrative designer, and an artist (specifically a concept artist). We liked to call this group the “DNA” of a character because it’s neat when an abbreviation works out and we all like to pretend producers don’t exist. (This is a joke. Producers are crucial to the process by which we make video games. But they’re often the “grown-up” in any room and get a bunch of shit for it ;P)

This meeting was something we called an “opportunity scoping” meeting. As in, we’re trying to usefully describe the dimensions of the opportunity represented by making a new character. Or again put more simply: the meeting in which we decide, at the highest level we find useful, what kind of a character we want to make.

That meeting went a little bit like this:

• Hey, we haven’t made a mid lane mage in a while.
• Yeah, and the last one we made was not really a mage at all. He was kind of weird and impossible to learn. (editor’s note: that would be Azir, and I feel comfortable shitting on his gameplay since it was I who designed his gameplay.)
• We’re working on Aurelion Sol though, right? He’s probably mid-lane?
• Yeah, but not exactly a normal mage either.
• And between Tahm Kench, Kindred, and ASol we’ve got a lot of non-human characters recently. We know players don’t attach to them as much and we should probably continue making more humanoid characters.
• Okay, so one option could be mid-lane, traditional mage, probably humanoid. Thoughts of power fantasy and place of origin?
• I think there’s gaps in our line-up of elemental mages. I mean we had Lissandra, who’s kind of a frost mage, and Brand, who’s a fire mage, but we don’t really have a wind mage (Janna is a support), a stone mage (Malphite’s a top lane tank or jungle tank), or an electricity mage (Kennen is a top lane fighter).
• Okay, cool, so humanoid elemental mage? Maybe? Anyone got any other pitches?
• Oh I wanted to say, for place of origin, there’s some story opportunities associated with Shurima. We’re planning [REDACTED]. Unless you mind working on two Shuriman mages, Daniel?
• (I did not mind at all)
• Cool! So let’s shoot for Shuriman humanoid, mid-lane compatible, mage, power fantasy related to an under-served element like stone or wind or electricity.

The meeting was a lot longer than that as you can imagine and dove into a bunch of other options, none of which really excited us. Then we all went away and each produced a little artifact to bring into the next meeting. I wrote a few exploratory paper-kits for differently themed elemental mages, our narrative designer wrote some story sketches and fragments of dialogue to find the voice and personality of this character, and our concept artist produced a gorgeous spread of five options.

You may be saying, wait, how can any of these people work on these things when they know nothing else of the character? And the answer is, you make a lot of guesses and you hold your ideas extremely lightly. This is all stuff we produce to be thrown away. You don’t sit down and write your first paper kit hoping for any of those abilities to actually ship.

Then you meet again a week later and compare notes. The art may change in response to an idea from the writer, the paper kit may change based on some element in the art, and so on.

That’s our first framework: the DNA process. This process is much more likely to result in characters that the individual crafts involved find interesting and worth making, because each craft is represented from day 1.

But secondly, I did not write these paper kits in a vacuum.

What a paper kit is for

Paper kits exist partly to communicate your gameplay ideas to other members of the DNA group, but they also exist for other game designers to look at and give you feedback. Saying “I want to make an electro mage, maybe with some kind of “attracts skillshots” passive, mostly damage focused with all of its CC potential locked behind an empowering ult”, you might paint a very similar image in the mind of an experienced character design on League of Legends, but laying it all out in a paper kit allows these other designers to critique specific ideas and patterns.

And this is where frameworks come in. Maybe not always explicitly, but always at least subconsciously. Here’s a few comments you might receive on our sample paper kit in the above inset from other designers:

• Oh, interesting. Are you sure you’re going to want to put all the CC into her ult? Her snowballed state in lane could become kind of oppressive with all that damage in her basic kit.
• Wow, she seems really teamfight focused. What’s the lane usage of her passive? Are you expecting people to electrify minions to arc to the enemy laner? Wouldn’t you have to electrify the enemy first? How long do you think that state should linger anyway?
• Are you sure the W turret should be destructible? That’s obvious counterplay, sure, but based on how much HP you give it you’ll probably end up in one of two failure modes: never worth attacking, so a noob trap, or always trivial to remove, so pretty much useless against ranged characters.
• Maybe the correct time cost model for the W is ward health, like Illaoi tentacles and Gangplank barrels? Still weak to ranged characters with an AA reset, but at least in lane it’s a knowable time investment.
• I’m worried about the defensive use case of W + E; if I read your kit correctly, she could always place a tesla coil over terrain and then E jump to it? Wouldn’t a really sophisticated player always hold the W + E combo for gank safety in lane? How do you incentivize offensive usage?

And so on. During early ideation (which is the phase the DNA process is part of in Riot’s ideation / pre-production / production / post-production cycle), a character designers spends a small amount of their time thinking about potential gameplay loops, character fantasies, or interesting kits and a much bigger part of their time discussing them with their fellow designers. Invariably both sides learn a bunch of things during this time, and more often than not other designers contribute ideas, skills they tried to make in the past, completely different character pitches, and so on. This is also where frameworks organically develop and improve.

But wait, why am I talking about frameworks? No one explicitly mentioned anything like that in the above examples, right? Well, not explicitly, no, but let’s dig into some of these example comments.

Deconstructing a comment into its component frameworks

“Are you sure you’re going to want to put all the CC into her ult?”

Maybe you’re an experienced League player and you asked yourself the same question. Maybe this doesn’t mean much to you. At any rate, the hypothetical designer asking this question is bringing a lot of assumptions into this:

• There is such a thing as CC, that is to say gameplay outputs that reduce the enemy player’s control of their character. (The literal expansion of the abbreviation is “crowd control”, and the reason it’s called that is lost to the ages.) (That is to say, I can’t be bothered to google what proto-MMO originated the language. It was already old hat at the time of Dark Age of Camelot.)
• Framework invoked: power budget. A character’s kit has an abstract amount of power budget, and that budget can be expressed in a bunch of ways, but primarily for an offensive character it’s split between damage and CC. An absence of CC in the basic kit therefore implies more damage than usual.
• Framework invoked: basic kit vs ult. In League of Legends, characters have 3 basic abilities and one ultimate. The ultimate becomes available at level 6 and is usually on a much longer cooldown (and, once upon a time when mana costs were real, used to cost a lot more mana). Ultimates are usually held back by sophisticated players for team fights or to secure kills that might otherwise slip through your fingers.
• Framework invoked: the laning phase. In League, the first 15 minutes or so of the game are expected to be taken up by the laning phase, where every character stays in their assigned lane and only interacts with the enemy/enemies in the same lane. From time to time the jungler comes to “gank”, but most of the time the lane is meant to be an even 1v1 (in top and mid) or 2v2 (in bot) matchup. Since we’re making a mid lane character, we assume a 1v1.
• Putting all of the above together, the observation becomes that due to the absence of CC in the character’s basic kit, the character will probably have a lot of damage in their basic kit. Damage, especially bursty damage, eventually becomes a game of break points. If you have enough to kill the enemy from their current health level in a single rotation, the enemy is essentially removed from lane. You cannot achieve the partial success state of hitting the enemy with some damage and some CC to stop them from responding in kind, and you cannot use CC to facilitate an easy escape. This means if you fall behind, you’re probably reduced to farming under turret and poking at your enemy with safe long range spells. If you get ahead on the other hand, your enemy may be removed from the lane. This all depends on the range and reliability of your damaging spells, of course, but in general, the heavier a champion is on pure damage, the swingier their lane. Look at Ezreal, Corki, and Lucian as mid-laners for instance.

Okay, oof, that’s a big example. I could go on for a while. But basically what we’re seeing in action here is a large set of theoretical frameworks that designers extracted, consciously or not, from examining the game as it is played by real players. Based on these frameworks, we can avoid easy mistakes and point towards improvements that are likely (but not guaranteed!) to make the character perform better, stand up better to repeated play, and function better in games of varying skill levels.

How this differs from rule of cool

In a rule of cool based environment, I might have made a very similar kit based purely on “electricity” tropes in video games. The difference is that other designers could not have looked at it and pointed out likely shortcomings and suggested fixes. We’d have to build and playtest it, and even then when problems occur (for instance, “she usually wins or loses her lane super hard with no in-between”), we wouldn’t know where to start looking. Without tools like frameworks, playtesters do or do not like your character simply because of how it’s tuned: that is to say, is it fun (aka over-tuned) or balanced (aka too weak).

In a rule of cool world, if you get lucky, you get an Octane or a Bloodhound and ship an amazing character. If you’re unlucky, you end up with a (pre-rework) Revenant or a Fuse–a truly super cool concept that’ll get players hyped only to completely fail to deliver on the fantasy.

And worst of all, as a designer in a purely rule of cool environment, you lack the tools to even guess at why your design for Revenant or Fuse didn’t work. You try a few abilities, chosen at random based on what seems cool to you, and if none of them work and you run out of time, you ship whatever’s least bad.

Or to use League examples, in the early days of the game when none of these frameworks were really mature and when the team shipped two characters every other week, you’d get rule of cool characters like Lee Sin, who absolutely stood the test of time and became one of the best and most beloved characters in the game, or you’d get characters like release Yorick, a complete mess of unintuitive math and irrelevant summons, which couldn’t be balanced into an acceptable state and ended up being nerfed into irrelevance until the team had time to do a rework.

I didn’t pick these randomly. They’re both the work of the same character designer, who is very very good at his craft indeed. But between the time pressure and the haphazard nature of “just make something cool we need to feed the content beast”, every character designed even by the best designer becomes a coin flip.

Moving towards frameworks

When I joined Respawn in 2020, I remember asking during my interviews what kinds of systems and frameworks they used to assess the design quality of their characters. I mostly got blank looks. Maybe the vocabulary was unfamiliar to them, I thought, and so I rephrased my question. The answer I got then, and the longer version of it that I got later was that they didn’t do that kind of thing. The vague feeling in the air was that this kind of thinking restricted creativity.

I actually think the opposite is the case. I believe constraints can breed creativity. But I also don’t think of these frameworks as constraints. I prefer to think of them as techniques that have worked before, starting points.

And it turned out there were frameworks at work at Respawn even then!

There’s always a framework

One of the first longer conversations about game design theory I had was with some of the OGs who’d been around for both Titanfalls. My question was, why do you think the multiplayer failed to find a sustainable population of players? Clearly both games were extremely well designed, players gave them rave reviews and so did professional reviewers (especially for Titanfall 2), and yet multiplayer numbers fell off quite sharply after not a lot of time. (And before you think it, no, I don’t think the release timing of TF2 had a lot to do with that. It surely torpedoed the game’s sales numbers, but there was a large enough playerbase for multiplayer, until there wasn’t.)

This is when I was introduced to the Brownian Motion framework.

It goes a little bit like this: the Titanfall games push mobility in a multiplayer FPS beyond the breaking point. While this is what attracts many players to them–after all, what other multiplayer game lets you double jump into wall run into double jump into grappling hook into wall run into double jump–it might ultimately also have been what doomed the games. See, when you lose an enemy around a corner, your brain constructs a mental map of the level with places where this enemy could have gone to by the time you get to the corner yourself. In most FPSes, this is a flat pancake shape that flows around geo. In Titanfall, it’s a rapidly growing bubble which isn’t slowed down by touching geo but rather accelerated by it; after all, wall running resets your double jump.

This means that when you lose an enemy, you LOSE them. The moment you no longer know where an enemy is, they might as well teleport to a random location on the map. This means that you’re entirely in a reactive mindset, responding to enemies as they pop up from unexpected angles in unexpected locations. By itself, that’s not a bad thing: it definitely checks a bunch of core skills in FPSes, such as reaction time, hand-eye coordination, precision and speed of aim, awareness of surroundings (to dash to the nearest bit of cover) etc. The problem arises from the fact that this is all you do. You never make a forward-looking plan, lay an ambush, anticipate an enemy pushing around a corner, or duck into a building because you can reasonably assume it’s safe territory. I have a whole ass design theory around monkey brain vs reptile brain, but this post is already way too long.

Anyway, take it as a given that being in a purely reactive mode is fun in the moment, but ultimately exhausting. It causes players to churn in a way where they can’t tell you why they churned. Usually the story goes something like “oh man, I loved Titanfall 2 multiplayer! I used to play it every day for a few weeks after release. And then one day I just stopped. I still don’t know why!” That’s burnout.

So that’s the Brownian motion framework. Nobody at Respawn called it a framework (and honestly that is on me for insisting on using pretentious words like frameworks), and when I prompted them for “how do you look at characters and decide which designs are likely to be successful”, nobody thought of this, but it quickly became super obvious that the Brownian motion framework was one of the most important razors for judging designs. It was really one of just two frameworks that leadership ever used to veto an idea before playtesting (the other being gun game vs character game, which prescribes that a player’s ability to use guns well should be the most important factor in deciding combat outcomes, not their ability to use character skills well).

Knowing this and understanding this helped me design a jetpack character that could simultaneously fulfill the fantasy of high mobility and not run into the excessive mobility issues from the Titanfalls. (And again, if you love the Titanfall games and still play them in multiplayer, chances are you love them BECAUSE of their mobility gameplay. When I say things like “excessive mobility leads to reactive gameplay which leads to burnout and churn”, imagine I’m saying “for most players” afterwards. There will always be people for whom a given design really really works, and that’s cool. Unfortunately, especially when making live service games, we have to be a little more mercenary and look at what design choices will yield a sustainable playerbase in the long term.)

Specifically, I attacked the root problems behind mobility instead of just reducing mobility until the jetpack didn’t feel good anymore. Without going into all the details, I made the jetpack very loud and noticeable so you could anticipate her coming, I made it leave a very pronounced lingering contrail in the air so you could figure out where she went after the fact, and I put a very noticeable weapon raise delay after you disengage the jetpack (and relied on the brilliant animation team at Respawn to make this delay feel good and active; look at how many different fidgets and gestures she has on disabling her jetpack) to discourage surprise attacks with the jetpack. While I’m not at Respawn anymore, I’ve been able to follow player reactions and balance follow-up, it seems she mostly landed in a good spot. That’s not because I’m a genius who can see the future (that part’s unrelated), but rather because the team developed these analytical tools, explained them to me, and I used them.

So if there’s always a framework, what’s your point?

Respawn had a great mobility framework, and a decent (if overly sensitive) gunplay vs character kit framework, but because these weren’t discussed in any way, I might have never learned of them had I not asked. Also, because the overall vibe was that it was bad to over-analyze and “constrain creativity”, many problematic design patterns did not have frameworks (for instance, there was no way to handle overall stealthiness, nor did the team have a lot of thoughts on selfish power vs selfless power, and those two gaps together doomed Revenant’s design), and even when there was a rudimentary framework, designers often did not have the confidence to call out problems before seeing them in gameplay, when it was often too late to do anything about them. For instance, Fuse was greenlit as the Explosives Expert, a very powerful and seemingly obvious fantasy that players really attached to. But a good understanding of the gun game vs character game framework should have made it clear from the start that promising big explosions and the damage that players would expect goes along with them was a non-starter.

Briefly, the gun game vs character game framework said that Apex is a gun game first and foremost; character abilities can change the context of a gunfight and give the player informational / positional advantages, but they should not by themselves decide the outcome of a fight. The problem with this is that a character whose two active abilities are both takes on “shoots a big explodey thing” creates expectations of damage done with those abilities, and when they turn out to be various types of “creates a damaging zone you probably don’t want to cross” and not something you can blow someone up with, you run into the “making promises your game can’t keep” problem.

Fuse’s original ult was a tactical nuke. It was a mortar that you fired at a destination, there’d be a big warning telegraph for enemies to leave the circle, and if they had not left by the time the nuke came down, they, uh, took enough damage that if you didn’t have a full healthbar and high level armor you were probably just dead.

This felt incredible for the Fuse player and was exactly what the power fantasy had promised. It was absolutely miserable for the people on the receiving end. Sometimes you just can’t leave a location. Sometimes the game puts so many other questions to you that you can’t also answer the “vacate the premises” question. Sometimes it was just asking the enemy “so which way would you like to die?”

The ability that shipped was kind of the inverse of the original ability. What shipped was a super cool and gameplay-appropriate circle of fire. Instead of making an ability that said “leave this place or else” we made one that said “btw you can’t leave this place or else”. It was designed such that dealing direct damage with the edge of the circle was nigh impossible, there were plenty of telegraphs and warnings and damage ramp up, and even when you did go through the flames, most of the power budget wasn’t even in the damage you took from the flames, but rather in the fact that you were seriously slowed when touching the flames and revealed while sitting in the ultimate: both ways to “change the context of the gunfight”, making a lot easier to kill the enemy with your gun.

But it wasn’t a big explosion that killed people. Neither was the tactical anymore.

And that fundamental mismatch between what the writing, art, and overall power fantasy promised (“blow people up”) and what the gameplay was able to deliver (“subtly alter enemy pathing”) was what we’d call thematically incohesive.

Let me linger on this point for just a second. I’m not saying the power fantasy was bad. Nor am I saying the gameplay was bad. Fuse has very specific use cases where he’s very powerful. What I’m saying is that power fantasy and actual gameplay are so different from one another that the overall character suffers for it.

And again, this is not the fault of any of the people who worked on Fuse! Everyone individually did great work, especially the character designer, whose first character this was, and who pivoted very quickly and competently from “big damaging explosion” to “cage enemies in”. Rather it was a failure of leadership. Since I was the lead game designer on the characters team, it was specifically my failure: my failure to catch the problem early enough for a useful pivot, and my failure to appropriately socialize and popularize our frameworks such that others could have caught this as well.

Conclusion

Embrace modernity. Have frameworks. Make them explicit, even if you never use the f-word. Discuss them often in your design team. Try and predict problems and successes and check your predictions against reality. Don’t forget to make a cool thing in the end. And maybe most importantly, don’t weaponize frameworks.

This point deserves a whole paragraph. It’s easy to use frameworks to shut down others’ ideas, especially if like me you can be loud and verbose and use terrifying multisyllabic words. It is always obvious when a framework is a useful razor to decide whether or not to try out a design. When it’s not obvious, default to trying it out. It’s called iterative game design for a reason. And when you think something was counter-indicated by a framework and it turns out it was fine, actually, update your frameworks.

And if someone tells you that they don’t want to restrain their creatives with all these rules, remind them that to break rules well one must first understand them well. Picasso didn’t start out painting fucked up shit. He learned and mastered the fundamentals and then painted a bunch of fucked up shit well.