Typo by Peter Seebach and Kevin Lynn [Comp04]

IFDB page: Typo!
Final placement: 19th place (of 36) in the 2004 Interactive Fiction Competition

The heart of Typo is the Frobozz Magic– er, Flavorplex Psychic Typo Error Correction System, which in turn is apparently a jazzed-up name for Cedric Knight’s mistype.h library extension. This innovation provides an extra level of player-friendliness by trying to catch unparseable misspellings and typographical errors and making a reasonable guess as to what the player intended, like so:

>push apadiing
[Flavorplex Psychic Typo Correction has divined that you want to "push padding"].
Your entire hand disappears momentarily into the padding. Now THAT'S soft!

This works, I’m guessing, by comparing the player’s input to a list of dictionary words, then using some set of algorithms to decide what in its list is the closest match to whatever the player typed. This is a cool idea, and frequently it works very well indeed. However, a utility like this is only as good as the dictionary it’s using, and if the game is underimplemented, typo correction can suddenly start wildly misinterpreting commands. Pushed further, its responses veer into the comical:

>x skyline
As you look out at the cityscape, your attention is drawn to a funny little pizza delivery car as it cruises slowly along the street.

>x car
[Flavorplex Psychic Typo Correction has divined that you want to "x cart"].
You can't see any such thing.

>[boy that can get irritating fast]
[Flavorplex Psychic Typo Correction has divined that you want to "buy that an get irritating east"].
You can't see any such thing.

The goofier moments reminded me of that Simpsons episode where Dolph makes a note on his Apple Newton to beat up Martin, which the PDA translates as “eat up Martha.” More seriously, the whole enterprise reminded me that everyone who writes software, including IF, must make decisions about interface. There’s been a trend in the last ten years or so, particularly noticeable (to me) in Microsoft products, towards interfaces that take a paternalistic attitude towards the user. “Don’t you worry your pretty little head about anything complicated,” they say, “I know what you really meant to do.” The problem, of course, is that at least half the time the “friendly” interface guesses wrong about what the user wants, and thus ends up thwarting good work rather than facilitating it.

I’ve no doubt that these sorts of interfaces arise due to a great deal of feedback from people who want their computers to be simpler to operate. However, for both naive and sophisticated users, it’s very frustrating when the computer thinks it knows better than you do. That’s why it’s critical that when you introduce a feature that overrides or embellishes literal user input, you provide a way to turn it off. In addition, such features are usually much better if they are customizable. A perfect example is the spell-checker whose dictionary can be expanded by the user — if I can teach the software more about what I want to do with it, I can better enable it to help correct only actual errors. So if anybody is thinking of using this system in another game, remember to implement every noun you mention and to give me some control over how and whether the typo system operates.

As for the game itself, it’s nothing too remarkable. Typo deploys the old reliable IF trick of literalizing some aspect of the medium, in this case the typo correction system. The PC is cast as a tester working for Flavorplex to iron out the bugs in its typo corrector. There’s one substantial puzzle, a Rube Goldberg device for which the PC receives a set of instructions, but which is constructed so straightforwardly that I never needed to consult them. There’s also one big plot twist, which in a more substantial game would move the action from prologue into the story proper, but which in this game serves only as an odd, abrupt, and unsatisfying ending. But Typo isn’t too interested in telling a story — instead, it just wants us to think about the implications of machines that make decisions on behalf of their users. For me, the game accomplished that goal.

Rating: 7.2

Janitor by Kevin Lynn and Peter Seebach as “Seebs” [Comp02]

IFDB page: Janitor
Final placement: 5th place (of 38) in the 2002 Interactive Fiction Competition

Note: Janitor is one of those games where you figure out what you’re supposed to be doing as you go along. It doesn’t take a terribly long time to reach this conclusion, but I’m going to talk about it, and that could quite reasonably be seen as a big spoiler.

We’re in the 8th year of the IF competition, and we seem to be at the point where we’re officially recycling stuff. Having just finished a game (Coffee Quest II) that often feels like a cut-rate knockoff of Michael Gentry’s Comp98 entry Little Blue Men, I discover Janitor, which more or less recapitulates the gimmick from Comp97’s Zero Sum Game, by Cody Sandifer. That is to say: an adventure has just happened, the score is at full, and the PC’s job is to go around unwinding all the accomplishments and setting things back to how they were.

Janitor seems more or less unaware of Zero Sum Game and its “game over” successors like Comp98’s Enlightenment and Comp99’s Spodgeville Murphy — for example, it doesn’t do anything entertaining when asked for the FULL score tally — but happily, it implements the unwinding idea in a more fun way than ZSG did. I was pretty turned off by ZSG‘s callousness towards PC, NPC, and player alike, and the way it motivated the PC to undo everything (the PC’s mother didn’t approve of killing or stealing) was undercut by the game’s taunting of the PC as a “mama’s boy” (or girl). Janitor‘s more sensible approach is to cast the PC as an employee of a text adventure company who must reset the game world so that the next player will find all the puzzles unsolved and the treasures in their original locations.

To accomplish this, the PC can use magical “access corridors” that connect to various rooms in the “proper” game scenario, along with a “mimesis disruptor” that’s good for quite a bit of nifty description-switching. The locations and objects are described with a great deal of humor, and the layout and puzzles are imaginative and clever. I wasn’t able to finish the game in the two hours allotted, which disappointed me, because it looked just about ready to open up into a new and fascinating level. I’d recommend this game to those who enjoy puzzles and IF metalevels, but I’d also recommend waiting to see if there’s a debugged post-comp release.

I say this because unfortunately, Janitor‘s implementation doesn’t quite match up to the wit of its premise and its writing. For instance, there’s a puzzle that can be solved over and over again, bringing the player’s score down to zero in a flash (which is the goal) without really accomplishing much. There’s a bucket that is supposedly full of water, but is also described as empty. What’s more, this bucket ends up being the game’s sack object, which really makes very little sense — why would I automatically store stuff in a bucket of dirty water?

In addition, there is a thin but even layer of punctuation and spelling problems in the prose, and the writing sometimes fails to mention certain critical actions that the game undertakes on behalf of the PC; I went a good long time thinking an object had disappeared, when in fact the game had actually moved it to my inventory without ever saying so. Perhaps most irritating of all, the first portion of the game doesn’t tell you what your goal is supposed to be, except for the rather vague explanation that it’s your first day as a janitor. This would be fine, except for the fact that it fails to respond well to your attempts to actually clean stuff — most such actions are either unimplemented or met with a discouraging message. I’d much rather have seen the game obligingly handle every request to clean something, and let the score mechanism clue the player in to the larger goal.

Also, while I certainly appreciated the effort put into the hints, I was frustrated by the fact that they didn’t go so far as to lay out all the necessary tasks. Consequently, I was left wandering around the game with 4 points out of 100, unable to progress to the next stage because I couldn’t figure out the last lousy task the game wanted me to perform. The hints’ reassurance that “our beta testers consistently beat the game” was the opposite of reassuring, helping me instead to feel stupid as well as aggravated.

I can most certainly appreciate the impulse to avoid giving the game away in the hints — I didn’t even include any hints or walkthrough in my entry last year, foolishly thinking that anybody would be able to stumble through it. Even this year, I wanted to just include hints and no walkthrough, but thought better of it. When somebody only has two hours to play my game, I want them to be able to see as much of it as they can, and it’s really not my place to police their “fun level” by making sure they do it the hard way.

Despite its problems, Janitor was interesting and amusing, and I wish I could have seen it through to its end in the space of two hours. A walkthrough or more explicit hints would have allowed me to do that. Instead, I’m left wondering what I missed, in more ways than one.

Rating: 8.2