Two readings of possible interest

The last couple of days have brought some interesting reads that weren’t announced on RAIF, so I’ll mention them here:

Trotting Krips’ review of Planetfall. I’ve never gotten around to playing this one myself.

Nick Montfort’s dissertation on nn, an IF development system he designed in the course of getting his doctorate at Penn. The dissertation runs to several hundred pages, so it’s not a light read, but I’d recommend a look to those interested in IF theory. Some of what he writes is fairly technical discussion of how his system works, and it’s difficult to judge its merits given that there aren’t any actual games written in it (as he admits himself); on the other hand, he also does a lot of theoretical definition of the different aspects of IF games.

One of the main points to be taken is his argument for the separation of the world model from the processor that describes this to the player. Some of our early discussions about this did affect suggestions I made to Graham about I7; nn goes much further, though, making it easy during play to switch viewpoint characters and to change the person of the narration, to change the tense in which things are narrated, to narrate events out of order or backwards, and to separate the viewpoint character from the person who is being given commands (something only a couple of IF games actually do — Nick cites “The Beetmonger’s Journey”; I seem to recall that “Bellclap” does a similar trick; but it’s not at all common). My feeling about nn so far is that it’s clever but also doesn’t go far enough: the sample output created tends to read as though still very mechanically generated, and it would be nice if the system went further in choosing which reports are most important on a given mood; collating related events together into a single description (which he does some of, but not as much as I’d like); etc. Still, this is hard stuff.

What I found more immediately useful to my own work were the analyses of narrative techniques in IF. Nick is very interested in looking at the different functions of text output by a work of IF, giving each of these different voices. Most obviously, he separates the narrator from the voice that handles meta tasks like saving and restoring and asking the player whether he really wants to quit; somewhat less obviously, he identifies a “suggester” voice, which is the source of clues and hints such as “You see something shiny in the corner” or “You could probably break down the north wall, if only you had something heavy enough to hit it with”. Generally IF authors tend to roll these things all together into the same collections of hand-written text; but what happens if we distinguish them? Might we want to write games in which the suggester can be set to be more or less explicit? or to react to the player’s progress so far? How much of this is worth modeling explicitly?

These are just a couple of points from a long text. Worth a look.

2 thoughts on “Two readings of possible interest”

  1. The idea of a the suggester voice being discreet from the naration is interesting. Most modern video games have given up the narrative voice, but they can’t seem to shake the suggestive voice. I’m seeing it more and more frequently personified in the protagonist’s ‘little buddy’. The bug in Okami, Daxter in Jack and Daxter, the dark princess thing in the new Zelda… seems to me like modern videogame heroes are taking too much instruction from things that ride them around. Anyway, in IF this sort of suggestive voice would be exactly the sort of thing I would throttle up for beginners and down for seasoned players.

  2. Yes, I’ve found myself thinking something very similar: there have been some games which offered variable-difficulty puzzles, but usually by changing what the player has to *do* to solve things. This is challenging to design (because it means that some objects may be necessary at one puzzle-difficulty level but red herrings at another) and it raises the amount of beta-testing the author has to arrange, because all the formats of all the puzzles have to be thoroughly tested. A much more accessible approach to the problem might well be to change how proactive and specific the suggester is. (But first we have to have conceptualized ‘the suggester’ as an aspect of the game.)

    So the next question is, how do we sensibly model the things the player might need to know in order to solve a puzzle, and how do we ramp up (or down) the amount that this gets revealed to the player?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: