- make sure disambiguation works sensibly
- provide implicit actions for things like opening doors, etc.
- synonyms for verbs and nouns in the game
- make sure the game can't get into an unwinnable state, or at least gives
an indication if it does
- check for text that might be seen out of order
- make sure there are in-game clues for important actions; give helpful
responses to "close but not right" actions
- think about alternative solutions for puzzles that the player might try
- check for any unresolved issues in the plot left hanging when the game
I have a mental checklist, which would have those things on if I
remembered them all. A couple more like:
- Make sure every mentioned object has a description.
- When complex actions are going on, make sure there are line breaks in
the correct place.
- For NPCs I try and list everything that the player might reasonably
ask about (and then a few more), and then implement those topics.
also more vague stuff like:
- Make sure the player always has some motivation and idea of what he
or she is trying to accomplish.
- Try and make sure no room is too cluttered or has too long a room
description. I always feel a bit intimidated where there is a room
with a long description and a load of complex objects to examine. It
can take a while of examining, and re-examining everything to start to
build up a mental picture of what the room is like. Not so good for
In that vein, ensure that line-breaks are consistent and
attractive. This is a little trickier in I7 that it was in I6,
for me, as in I6 I understood all the conventions, e.g., the
describe property needed a newline before and after any text I
printed, while before-like routines only needed one afterwards.
> - make sure there are in-game clues for important actions; give helpful
> responses to "close but not right" actions
> - think about alternative solutions for puzzles that the player might try
> - check for any unresolved issues in the plot left hanging when the game
That last one depends on the genre.
We don't necessarily discriminate. We simply exclude certain
types of people. --Colonel Gerald Wellman
Not sure I agree ... but then, I have a general tendency to write long, so
YMMV. For me, a room with a one-sentence description and only one object in
the room that can be examined feels half-hearted and inept. For me,
"immersion" means there's lots of interesting stuff to do.
The advice of Gertrude Stein may be apropos. In "How to Write" (if you can
find a copy, don't bother to try to read it; it's impenetrable), she wrote
at least one lightning-bolt sentence: "A sentence is not emotional a
paragraph is." This tells me that emotion -- or call it dramatic affect --
lies in the juxtaposition of two or more sentences. So a one-sentence room
description has, by definition, what the psychologists call flattened
When I was working on my first attempted game, I would have liked to
have a comprehensive list of things that I might not have thought of on
my own, but which a lot of people here think is essential for a decent
Unfortunately, the only example I can think of right now is "describe
Sure, I'd say excess in either direction is bad. I wouldn't advocate
one-sentence rooms. Where to draw the line is subjective, but it's
preferable when there are less important objects in a room than can be
remembered in one read through of the description.
I guess it could be deliberate in some cases, but in general it can be
frustrating to end a game and have no resolution for things brought up in
the game ...
A quote from Dan Shiovitz about Waldo's Pie (2005):
I'm sure the author knows all the missing pieces, but just because the PC
has been hit by a memory scrambler is no reason not to give the player
answers to some of the important questions by the end of the game ... so,
yeah, fun but kind of frustrating.
That's a good one.
- make sure any major changes to the PC or the world result in updated
- check for incogruous default messages ...
(from a review of "Jane" (2002) by Paul O'Brian):
At its worst moments, the clash between the intense action of
the story and the standard Inform library responses evoked
by my actions was outright comical, completely defeating the
John's lost in his mind again. "You ARE nothing!" he shouts
again. He steps forward quickly and shoves you back, causing
you to stumble to keep your balance.
"You're useless! You're so %$@# useless!"
That would be less than courteous.
I agree. Even if you only have one room in your game, you shouldn't put
every object in the room description, or it will overwhelm the player
immediately. If you want to have five pieces of furniture, each with
several named parts and many small items on/in them, that's fine, but
you need to condense the descriptions down for the main room
description. If there's a lot of stuff in the room, describe each
individual item sitting on the table only after examining the table.
A related problem that I've encountered (and one for which I have no
good solution) is the issue of how many objects in a given room should
be described (and how many need to be implemented as more than scenery)
in order to keep the player immersed in the setting. There are very,
very few real-life scenarios that have few enough things in them to
feasibly implement with anywhere near full realism.
A realistically implemented set of silverware drawers would have
teaspoons, soup spoons, forks, butter knives, steak knives, spatulas, a
ladle, a whisk, serving spoons, large wooden stirring spoons, a cork
screw, at least a dozen other easily identifiable cooking implements,
and a couple random doodads of unknown purpose. Can you imagine trying
to implement an entire kitchen's worth?
Obviously you don't implement the whole kitchen, but where do you stop?
Do you describe the contents of each drawer/cupboard broadly
("silverware," "cups," or "plates") and artificially limit interaction
with them? Sometimes that can work, if you can truly assure that no
reasonable person would want anything out of the silverware drawer in
the course of your game. This is perfectly acceptable in some of the
more plot-driven games, but it's hard to imagine a puzzlefest where
nothing in that drawer would seem useful to a player.
My favorite technique is to describe and fully implement a "significant"
number of items in a room, enough to give the player's subconscious mind
the impression that the room is completely full, but at the same time
avoid mentioning the unimplemented aspects so as not to give the same
impression to the player's conscious mind. So the trick would be to
describe and implement (a stack of cups isn't too bad to deal with)
enough of the dishes in the kitchen cabinets and simply not mention the
silverware drawer at all. That way, when the player imagines the room
in his head, it probably looks like some kitchen they've actually seen,
but when it comes time to decide what to do, they think about the stack
of cups and not the silverware drawer.
The problem is that I don't yet have the knack for finding that balance.
Anyone have any tips on how to decide what to leave out?
> Slightly related:
> - check for incogruous default messages ...
It's probably best to replace all the default messages, unless you're
consciously imitating Infocom's style (hello, Curses). The standard
library responses are not vanilla neutral; they imply and convey a
certain attitude, which most modern works of IF do not share.
- Implement verbs and adjectives, not just nouns, used in the text.
- Get beta testers and listen to them. If you disagree with what they
say, chances are you're wrong and they're right.