Thread by @Delafina777

So one of the weirdest things about writing spoken dialogue for games is realizing that you have to put some noise back *in.* (Thread)

If you listen to how human beings talk, on average our noise:signal ratio is pretty high. There's a lot of "um"s and verbal filler, and English speakers can't usually go more than a few sentences without using an idiom or other preset phrase.

In an average conversation, you could probably finish a lot of the other person's sentences, even if you don't know them that well.

In TV or movie dialogue, there's this constant tension between realism and beauty/usefulness/etc. In most movies and shows, we want the characters we're watching to speak believably, but we also want their speech to be better than reality.

So it's this fine line between not making it TOO elevated/polished, and still making it more interesting and pleasurable than the vast majority of real life speech.

Writing dialogue for games is an even weirder balance, because you don't have the tacit agreement from the audience to be passive watchers/listeners, which means the audience is far more impatient.

When I was working on Kinect (the Xbox motion-tracker device, which usually had you stand to play), I sat in on some playtesting with little kids. If a character talked to them (or characters talked to each other) for more than (IIRC) 20 seconds or so, they'd sit down.

They were kids, so it was this very natural, honest feedback:

they'd go out of game-playing mode and into tv-watching mode.

When we talked about it afterward--well, we were game-players ourselves, as well as game-makers, so we knew our own impatience.

But this really pointed up how little passivity and inactivity players will tolerate while playing a game.

And that's not to say that you can't have more than 20 seconds of dialogue ever--cut scenes are obviously still a thing.

But it seems like after a very short period of time with nothing to do but watch/listen, players mentally switch modes.

What watching those kids play with Kinect, where they had to be standing, did was make that mode-switch physically visible.

So, I think everyone who writes for games has some instinctive sense of that--you have to keep periods in which a player isn't *doing* anything and a character is talking very short, unless you've communicated (like with a cut scene) that they should go into watching mode.

On top of that, there are almost always budget constraints on how many VO lines you can have, and often technical constraints (like character limits) on how long lines can be.

So the pressure is

SHORT SHORT SHORT

every line has to do multiple things:

-communicate information about what to do or how to play
-be entertaining
-develop character
-develop setting

Because players will get impatient and start trying to do stuff, or stop listening, very, very quickly.

So you end up trying to make every line EXTREMELY pithy.

Obviously in general you try to work in exposition gracefully and as invisibly as possible--a phrase of explanation in an instruction, through ambient dialogue, and--if you're not too silo-ed, through visuals, through the environment itself.

There are ways to cheat, too--if you have to drop significant exposition and there's no other way to do it other than having a character talk at you, you make the PC walk/ride/etc somewhere so they're *doing* something, and then tell them while they're actively traveling.

It's basically: "give them something to do that uses their controller but is relatively mindless so that they're not just standing around, but still have nothing better to do at the moment than listen."

And man oh man, if you think players don't *listen,* how much more do they not *read.*

Especially on a console. There are all kinds of issues--you don't know how far away they're sitting, so text might be too small, and people don't *read* on TVs the way they do on monitors...

So when it comes to writing text that isn't spoken, it has to be *extraordinarily* pithy. Because they're not reading more than a couple sentences, at least on a console, and are going to be annoyed that they have to do it unless they're taking the initiative to seek it out.

So you get into this hyper-polish mode. Every line, you're looking for ANYTHING you can take out. Every word has to pull its own weight and then some. Cut ALL the fat.

And that super-concentrated approach does actually work for text.

Surprisingly, it didn't for spoken dialogue.

What happened was that the first people to play it--who were people *working* on the game, so already familiar with it, couldn't process it.

Simple instructions, sure. But not plot, not the *reasons* characters were doing things.

It was too concentrated. Even though they were listening, and even though these were people who already knew the plot, they still just... didn't catch it.

Pure signal (or at least, as close as we could get to it) didn't work.

We had to put noise back in.

I spent a lot of time wondering about that. And then I was out to dinner with some friends at a restaurant that was kind of noisy and I realized why so much of our every day communication is noise.

And why so much of it is set phrases.

We're rarely fully focused on listening. Our attention flickers in and out. This isn't because we don't care or aren't *trying* to pay attention or anything like that.

It's because the world is noisy. It's because even if we don't feel like we're paying attention to them, we're still aware of physical sensations. It's because a truck goes by and we can't hear a couple words.

And by and large, that doesn't matter, because the way we speak is designed to compensate for that oscillating attention.

If I were talking to you, and started with "And by and large," your brain would know where that phrase was going and could look away.

The rhythm of how we speak, the phrases we use, all that--it's designed to give our brains breaks from listening in order to process, and to usually make it clear when novel/important content is coming so we don't miss it.

The noise is important to being able to process the signal.