I’ve written a few times already about my new, novelette-length choice-based story built in Undum. The project began life as technical experiment – a quick attempt (a bit like this one) to “do” a text-game as a multiple choice adventure. The concept was simple: the game would have locations, and objects, but streamline the usual breadth of Interactive Fiction’s parser down to just the choices that mattered for the story.
It didn’t work and I had to change the design. But I learnt a lot in the process.
The game, No Space to Breathe, starts off in a corridor, outside a room, with a guard. (Guess the source game, if you like.) You had four options: look at your possessions, examine the guard, talk to the guard, open the door and go inside. Choosing to talk would bring up more choices; looking at what you’re carrying would bring up the option to read one or other of two documents; and alongside was the option to go back up a level (“Do something else”).
And so on. Opening doors, picking things up, using them, and all the while, the option to examine things, over and over again.
I wrote about a quarter of the game this way before the – surely obvious conclusion – hit me, that it was a terrible, terrible structure for a game. It played horrendously. The problems were manifold:
- If there was an option I wanted to take, it wasn’t always clear what submenu to find it on
- If there wasn’t an option available, it felt weirdly like there was a gap
- If I took an option (examine the guard), then went back up to the same menu, the same choice was there again, as if I’d never taken it
- If an option was listed but turned out to be unimportant, it felt strange that it had been there – and doubly strange that it was still there later
- It was easy to forget what menu level you were on, and indeed, you didn’t really care. The result was, it became difficult to move the game forward
One could group all these problems together under a single banner: the interface made the game look static and stupid, and like it had no idea what was going on.
But the effect is even worse than that. The problem is, during the first few moves of a game when a player is getting themselves orientated, they tend not think “this game is stupid”. They’re more likely to think that there’s system, but they’re just not getting it. An interface like this one – an unintelligent, static one, with all the options listed all the time, is more than clunky; it’s actively misleading.
Because the choices provided by a game – or by any UI system at all – are more than just the sum of their names and icons and positioning. They’re also affected by their content – which is to say, the meaning the player assumes a choice to have will depend, at least in part, on the other choices offered alongside.
Which is to say, people expect a choice to be a choice.
Maybe this is obvious. It’s certainly true that any choices you might think of in real life are selections between vying alternatives. From buying a chocolate bar to picking a dentist, or selecting which tool to use on a filling, whenever we think of a choice we think of alternatives that exist on a par with one another.
The problem with my text-game-as-clicking-buttons was that the choices, placed side-by-side, became suddenly, and spontaneously, absurd.
The king rolls the die in his hand, thoughtfully.
- Talk to the king
- Examine my possessions
- Pet the dog
So. I rewrote the game, trying to bring all the choices into line with one another. Gone was the ability to click a link more than once; gone was the ability to research freely. Gone was unchecked navigation, conversations that could be dipped in and out of at will. The result? Not just a less confusing game. Not just a less broken game.
What I got in return was a game with a stronger sense of place.
Imagine the scene:
- Drink the coffee
- Read the newspaper
- Listen to the radio
Compare that with:
- Drink the coffee
- Go to the garden
- Check your email
There’s nothing wrong with either set at first glance, right? But of the two, the first creates a sense of place; the choices bind each other together into a general sense of morning-breakfastness. They add up to more than their sum: despite the player only being able to click one, having the three on offer actually improves the story.
By contrast, the second set has no inter-association. It’s hard to remember the choices once you’ve read them; they don’t fit together. They don’t characterise the protagonist, his location or state of mind. (Or worse, maybe you find yourself creating explanations when reading them. Perhaps the character is a procrastinating writer, or a job candidate, waiting for the phone to ring?)
Now, in case you think this is a problem peculiar to text-games, have a think about LA Noire. I’ve written before about the weird interrogation mechanic – three choices: Truth, Doubt or Lie. The game tells me I should decide based on the body-language and facial expressions of the NPC – except if I want to call a Lie, in which case I’m supposed to decide on the basis of the evidence I’ve collected.
The reason I get this wrong, and forget how the game works, is exactly one of context. Truth and Doubt are alternatives designed for me to decide what I think about the character. Truth and Lie are there so I can sift the evidence. The distinctions are pretty clear, but as soon as the three are available at the same time I start wondering, again, what the difference between Doubt and Lie is, and a smooth, simplistic interface becomes cluttered and confusing.
The choices we produce in a game are the player’s only way in to affect the protagonist, but they’re also a crucial way for things to come out too. Choices represent the bits of the game we’re putting in the player’s mind for them to turn over and think about. However good your graphics engine or your sizzling prose, the choices themselves will be the most real part of the experience.
And like any other part of a piece of writing, the choices represent an opportunity to add value or damage your work. So whatever we do, we shouldn’t let them be decided by algorithm or a static tree. We need to craft them as carefully as we would craft the words themselves.
(Postscript: The only alternative is to make our choices invisible, the way a console joystick does, the way parser IF does, the way keyboard shortcuts in a word-processor do. Richness of experience goes up, accessibility goes down.)