Average IF command lengths

In another thread I said: “80% of commands are single words (if not single letters), and 80% of the rest are two-word command phrases.”

Those number were, how shall we say, colonically extrofitted. therealeasterbunny called me on it, so I’ve done some counting.

All of these numbers are from my beta-testing transcripts.

The first entry for each game (“with one or two letters”) is a subset of the “one word” entry.

Note that Heliopause and Dual Transform have very different proportions from Dreamhold and Wallpaper. Heliopause and Dual Transform are both one-room(-ish) games. They each have a notion of locale, but navigation is by two-word commands in each case, with no one-or-two-letter abbreviations.

Conclusion: my original numbers were too high, but not very high (for games with traditional map navigation). Between 50% and 70% of commands are one word. About 75% of the rest are two words.

* Delightful Wallpaper
3005 input lines
with one or two letters: 66.1 percent
with 1 words: 69.5 percent
with 2 words: 20.8 percent
with 3 words: 0.6 percent
with 4 words: 7.2 percent

* The Dreamhold:
10444 input lines
with one or two letters: 43.2 percent
with 1 words: 53.0 percent
with 2 words: 31.1 percent
with 3 words: 6.0 percent
with 4 words: 4.8 percent

* Heliopause:
1107 input lines
with one or two letters: 19.1 percent
with 1 words: 36.6 percent
with 2 words: 49.2 percent
with 3 words: 6.1 percent
with 4 words: 3.0 percent

* Dual Transform:
2687 input lines
with one or two letters: 14.4 percent
with 1 words: 20.3 percent
with 2 words: 71.5 percent
with 3 words: 2.1 percent
with 4 words: 4.2 percent

I wonder how much this is affected by your particular style, and in particular whether any of your games required any commands of more than two words.

It’d be interesting to see what the results were for, say, Hoosegow; in its walkthrough, 36 out of 105 commands have more than two words. Several of those commands involve two-word nouns that can probably be cut down for lazy typists (“x right” rather than “x right whatever”), but some can’t. In particular, dealing with NPCs seems to involve a lot of four-word commands; “ask X about Y,” “show X to Y.” There are also a fair number of actions involving two things (like “Throw x at y,” though that in fact isn’t one of them). And in the latter category I did wind up retrying a lot of those actions, so my proportion of multi-word commands may have been inflated.

Of course Hoosegow was another mostly one-room game, so you don’t get the navigational commands.

As noted elsewhere, looking at a walkthrough is fairly useless, because nobody plays that way. (Except the game author, constructing a walkthrough.) You have to look at actual play transcripts.

Oh, I didn’t mean to suggest that we could conclude that one-third of the commands entered in Hoosegow would have more than two words. Just using the walkthrough to note that you did need to use a lot of multi-word commands to finish. (Well, it doesn’t prove that, but my recollection is that a lot of the commands had to be multi-word.) Just wondering how different kinds of game would affect the count. For instance, I don’t think the three of those games I’ve played (Dreamhold, Hoist, Dual) had NPCs, or maybe didn’t have traditional NPCs, so I wasn’t SHOWing anything TO anyone. My idea was just that it’d be interesting to look at stats from other types of games to see what they looked like.

Anyway, I’m trying to get hold of a Hoosegow transcript; if I do I’ll come up with some real stats. (If anyone has a transcript and wants to do the count themselves, that would be particularly awesome.)

I have linked the online version of Raising the Flag on Mount Yo Momma to a modified Parchment that saves the transcripts as people play. The sample size means that nothing of this is in any way statistically significant – these are based on 416 commands over 16 played games (on average 26 turns/game), none to completion.

  • with one or two letters: 24.0 percent
  • with 1 words: 40.9 percent
  • with 2 words: 42.8 percent
  • with 3 words: 12.0 percent
  • with 4 words: 3.1 percent
  • more than 4 words: 1.2 percent

The Parchment data collecting demo has slightly more data, 681 commands over 23 played games. Roughly one third is from a single playing of Escapade! to completion, 261 turns. The rest are from playing Yo Momma, most of them presumably just trying out the system, but the numbers are still quite close to the other data set. (Note that Escapade! is a one-room game so there are much fewer one-character compass directions in this set.)

  • with one or two letters: 11.6 percent
  • with 1 words: 39.9 percent
  • with 2 words: 43.2 percent
  • with 3 words: 9.0 percent
  • with 4 words: 7.0 percent
  • more than 4 words: 0.9 percent

I’m not sure how hard this would be to automate, but I’d be interested to see what percentage of commands were navigational – the compass directions and in/out, primarily.

Thanks, Juhana – Raising the Flag was a game I was interested in, because it’s got a lot of NPCs but also a fair amount of movement.

Do you think the three-word commands were mostly “TALK TO NPC” or “LOOK UNDER/BEHIND THING”? Or something else?

BTW, I don’t doubt zarf’s general point that most commands are two words or less, unless you’re playing BOWL DROP TOY SLEEP PLAY or something. (And even then we have the caveat that you can’t use a walkthrough for these statistics.) I’m just curious.

That’s right, about 75% of all three-word commands were TALK TO [someone]. (Suprisingly no LOOK UNDER/BEHIND - apparently these players used SEARCH instead.)