Out of world (extradiegetic) interactivity

I’m curious what others think about games that try to adapt the story to the style and attitude of the player rather than (or as well as) more “logical” consequences of the player character’s actions.

For example: (and this is just totally ad lib) “you” find a stack of books. There are some that deal with caves, urban exploration, speleology, and the legendary subterranean world of Agarttha, and another stack with books about surveillance, the panopticon and miniature cameras. Oh, and there’s a spiral staircase going down, down, down. There’s a metal gate at the top you need to get past first. But what’s down there?

Now logically, in the real world, whether you stop to read the books - and which books you read - will not actually change what’s at the bottom of those stairs. But in a game, it might.

if you’re analytically minded, this is about interactivity on two levels: there’s stuff the character needs to do in her own world - find a key to get past the gate, and other stuff that happens where this a direct causality between the player character’s actions and later events. This would be the diegetic interactivity to a narratologist, I suppose.

The other level also has a kind of causality, but works in another way. The story changes, but not in a way so that those change can rationally be seen as caused by the actions of the character. Whether there’s a high tech surveillance room or a large underground maze down there is in caused by what the player character reads, but that causality is not in the world (i.e. this is extradiegetic interactivity).

Also I guess these two levels correspond to the player/character duality. One level is responding to what the character does, the other is trying to adapt the game directly to what the player seems to be interested in.

I’m not sure how much of this is too much. I’m afraid it will be too meta, and turn the playing of the game into a game in itself, if you get what I mean. And will that mechanism control how the player approaches the game, rather than be controlled by how the player behaves? And what does it mean for replayability? Will returning players skip the books (since the player already knows what’s in them - reading them again is just boring) only to find that this time the paranoid watchman down there is gone and replaced with a battle droid (reserved for players who don’t read books at all)?

I guess I’m wondering… Is this something you want? And what experiences (if any) do you have trying to implement things like this?

Also, this has been something I meant to ask for a while, and similar stuff is now talked about in other threads, but I don’t want to hijack those threads. To clarify though, I’m not talking about generating content on the fly. And realistically, maybe it’s not about replacing large chunks of the game, but perhaps trying to adjust how the game is described, fiddling with the “dark and moody - tongue in cheek” slider, things like that.

I’d like it if it were awesome. I’d like it less if it weren’t.

As with any other abstract, it’s all about the individual work. I think anything can be awesome, with sufficient awesome applied.

A few examples of vaguely similar things:

Metamorphoses tries to track whether the player uses brute force or cleverness in various contexts, and adjusts things slightly in response to that fact. To the best of my knowledge hardly anyone realized this during play and it’s sufficiently unimportant that I’ve largely forgotten the details of what it changed. But in theory it was trying to extrapolate things about the player character this way. Daniel Freas’ The Erudition Chamber overtly and explicitly tests for player puzzle-solving style, and this is a large part of the point of the game.

Both Blue Lacuna and City of Secrets attempt to do some guessing of the player’s skill level to self-adjust how hard things should be. I am not sure how successful this was in either case; I had some complaints about this from players on CoS, and Blue Lacuna incorrectly guessed that I was a novice player when I tried it, probably because I was more interested in the story-relevant actions than in trying to solve puzzles immediately.

There are a handful of other games that do things like guess the player’s gender and/or sexual orientation based on what you choose to interact with.

My observation is that often the heavy lifting work of creating all this variation is wasted if it’s done subtly: the more you’re adapting the game to the player, the more extra content you have to make, but will the player appreciate it if he never realizes that that’s what happened? Conversely, if the switches are really overt, then they can come off feeling not much different from just having a menu at the beginning of the game that says “Are you male or female? > Puzzles or no puzzles? >” and so on.

I think I understand. The first book the player reads might determine what happens to be at the bottom of the stairs. This is even more extradiegetic, or out-of-world or whatever, than the Infocom game that assigned the player character’s gender based on what bathroom the player chose to enter, because the books that control the content downstairs is more a matter of theme and meaning.

I did something less extradiegetic than the example, but probably still perceived as such, because the causal relationship would only be apparent to someone who played “Tree and Star” twice and chose different main branches each time, which is almost impossible to do with the state that the released Andromeda Comp version of that game is in.

  1. This is already pretty common in CRPGs with skill, D&D-type stats, and (crude representations of) ethical status. (The player likes sneaking-and-treachery type plots, so they make a thief character and the game gives them thiefy quests). Aesthetics/tone/story element stuff is more interesting, probably because it’s a whole lot harder to do.

  2. It’s an assload of of work to do it non-trivially.

  3. It’s sort of boring to do it along a single axis; also, this puts more pressure on your assessment-mechanic to get it right. (At the moment, I’m idly fiddling with a ‘here are five aspects of the game the player could be interested in; the top two get foregrounded’ approach.)

  4. It’s better to make it somewhat flexible, so the player can switch focus if they don’t like where something’s taking them. (Also, it’s boring and anti-story to have to commit to a particular character build early on.)

  5. Inline notation for variable text is invaluable. Trying to do this in a system that doesn’t have powerful variable-text tools will be like pulling teeth.

Players are already trying to guess the game’s agenda based on what it presents, so when the game tries to guess the player’s agenda, there’s a risk of goofy feedback loops.

A shallow example of this, which I run into in all sorts of exploration-based games: I usually try to explore short side-alleys before continuing down the main road. (Whether this refers to physical landscape or the plot diagram.) This often means not doing the obvious thing first. But sometimes a game wants me to spend some time poking around in side alleys, so it shows them to me first, and de-emphasizes the main road. I wind up second-guessing wrongly.

This rarely causes a serious play-experience disaster, as long as the game has its fundamentals in order. (E.g., it doesn’t throw me through a one-way portal into chapter 3 until I’ve clearly indicated I’m satisfied with chapter 2.) But if you’re conditioning big chunks of the game on this sort of thing, second-guessing at cross-purposes can be a problem.

That’s a really good point, Zarf. I think the most important extradiagetic choice that players make is which game to play. Once they’ve gotten that far, authors should give them what they do best.