Mechanics spread over too much game


This is an essay motivated by the question of why so many games based on movies are no good.

First, some terminology. By game mechanic, I mean something you do in the game together with all the aspects of the game that support it: input, camera, physics, graphics, reward structure, level design. For example, racing games are usually based around a mechanic in which you drive a vehicle on a track: you can steer, accelerate and brake, the camera stays behind and slightly above your vehicle, always pointing in the direction you’re facing so that it shows the track ahead. The physics engine handles things like momentum and grip. By core mechanic I mean the game mechanic that you spend the majority of your time doing.

These terms originate in board and tabletop game design—Wikipedia has a decent article—but they extend in obvious ways to video games.

In many of the best video games there’s a single core mechanic that is done really well and the whole game is built around that. Few games manage to do more than one mechanic well; usually subsidiary mechanics suffer. Three examples: Ico has two mechanics, one based on walking and climbing in order to find a path that both you and Yorda can travel; the other involves hitting things with a stick. The walking/climbing mechanic is done superbly well, but the fighting gets dull and repetitive. Prince of Persia: The Sands of Time is similar to Ico: climbing and exploring is great, fighting is really tedious. No amount of graphics can save it. Ratchet & Clank ambitiously attempts three: platforming, shooting stuff with gadgets, and racing. The shooting game is fun, the platforming is ok and the racing is really dreadful.

Racing is a particular problem. Developers who surely realize just how hard it is to make a proper, stand-alone racing game, seem to think they can dash one off as an afterthought in the middle of another game. (Ratchet & Clank and Star Fox Adventures are particular offenders; the Legend of Zelda games are also prone to have unpleasant racing minigames, but are more excusable because the races can usually be skipped).

Why is it so hard to create a successful game mechanic? Because it requires so much attention to detail. The input has to be responsive, every action needs to come with feedback, both aural and visual (and force-feedback if your controller supports it), and actions and animations need to feel so natural that you can forget that you’re just pushing buttons on a controller. There’s the structural aspect: you’ve got to have a goal, the goal has to be clear to the player, each step has to be appropriately rewarded and deviations should be appropriately discouraged. There are graphical aspects too: the camera has to be pointing in the right direction, and the important aspects of the world have to be emphasized. And of course all of these things have to be engaging and fun.

I think the only way to get all of these things right is to try things out, play them, identify what’s succeeding and what’s not, and repeat.

You might imagine that you can bypass all this work by copying a successful mechanic from another game, but it isn’t as easy as it seems, because unless you’ve personally gone through the testing/refining process you won’t necessarily be able to say just what it is about that game that makes it successful—the sheer number of awful clones out there should make designers realize that it’s not easy to copy success. And anyway, unless you’re only making an exact clone (and even that often fails—there have been poor versions of games as simple and apparently bomb-proof as Tetris) you’re going to want to introduce unique aspects to the gameplay, and that means looking carefully at and adjusting all of these aspects. It’s naïve to think, “oh, it’s just another platform game, so it’ll play the same”. Something as simple as a change in the character graphic can radically alter the gameplay—perhaps the jumping animation that looked fine on a cartoony character now looks absurd on a more realistic character, so you need to slow it down, but now the feedback isn’t rapid enough for twitchy reactions, so you need to make the platforms a bit bigger and the jumps easier. Small differences can have knock-on effects that are hard to predict. There’s really no substitute for repeated testing and refining.

It’s tempting for a producer to think along the lines “this game has ten walking levels and only one driving level, so we should invest ten times the designer/programmer time in the walking part of the game as the driving part”. You might be able to divide up the artist and level designer time like that, but you can’t divide the programmer time because it takes the same amount of time to implement a good mechanic regardless of how much time the player is going to spend playing it. If you fall for this fallacy of proportionality, you’re going to be tempted to add lots of game mechanics and spend only a little time on each. But then they are pretty much guaranteed to suck. A really good mechanism pays off the investment in its development because you can keep adding new levels or even new games using the same mechanism and they will still be fun to play.

In a long-running game series it’s possible to add more decent mechanics over time: for the new game you can keep the successful mechanic from the old game, more or less unchanged, and use your resources try to add new ones. Sometimes this works very well: the Legend of Zelda series has managed to acquire several superb gameplay mechanics, for example the “deku scrub” gameplay in Majora’s Mask. Sometimes it doesn’t. Some Zelda games contain many dull mini-games; the other playable characters in Majora’s Mask weren’t nearly so fun as the deku scrub. A lot of Mario games implement swimming but it has never worked particularly well. The Grand Theft Auto series extended its excellent driving mechanic with a much less successful walking and shooting mechanic.

This is a particular problem for film tie-ins. The game designer looks at the film and sees sword fighting, a car chase, some shooting of guns, some running around on platforms, and all the other things that make up the film, and decides that the game will feature all of these. But that’s absurdly ambitious: it is the work of a complete game project just to make a good car chase game. You can’t do that and all the other things unless you cut corners. Some of the mechanics are going to be bad and very possibly all of them will be.

The most successful film adaptations are those that have stuck to a single mechanic and done a good job. GoldenEye 007 looks at the whole panoply of actions in the James Bond film and discards everything that can’t be implemented using the first-person shooter mechanic. Star Wars: Rogue Squadron picks spaceship combat from the movies and ignores the rest. But you just know that somewhere there’s a game producer thinking that it would have been improved by alternating the spaceship missions with lightsaber combat and devoting half the developer effort to each half of the gameplay. He doesn’t realise that the most likely outcome would have been a game that was only half as good.

I feel like I’m stating the obvious here, but it can’t be that obvious or there wouldn’t be so many games that do too many things and end up doing none of them particularly well.