Getting games into unsolvable states

12 views
Skip to first unread message

Gareth Rees

unread,
Nov 27, 1994, 10:06:46 AM11/27/94
to
Some minor spoilers for "Curses" in the following:

I'd like to canvas the opinions of the people who post to this newsgroup
about the pros and cons of letting games get into states from which they
can't be solved.

"Curses" seems to be a well-respected game, but it's ludicrously easy to
get the game into an unsolvable state without realising it (e.g. by
reading the poetry before getting the attic key, or visiting the Tarot
cards in the wrong order).

It seems to me that making it easy to get the game into an unsolvable
state is running the risk of causing the player to lose interest
(because they don't seem to be getting anywhere, or because they
discover at some point near the end of the game that all their save
files are worthless and they have to start all over again), and should
be avoided as much as possible.

But on the other hand, if you try to avoid this possibility then there
are big restrictions on the kind of puzzles you can design - no puzzles
that have time limits - NPCs can't get bored and wander away -
opportunities have to keep re-presenting themselves - and so on.

Any thoughts? What's a good balance to strike?

--
Gareth Rees

Damien P. Neil

unread,
Nov 27, 1994, 11:46:19 AM11/27/94
to
In article <3ba7a6$6...@lyra.csx.cam.ac.uk>,

Gareth Rees <gd...@cl.cam.ac.uk> wrote:
>Some minor spoilers for "Curses" in the following:
>
>I'd like to canvas the opinions of the people who post to this newsgroup
>about the pros and cons of letting games get into states from which they
>can't be solved.

First of all, killing the player is, in a way, the most basic method of
placing the game in an unsolvable state. If you are truly serious about
removing dead ends, you have to remove death. In my opinion, however,
player deaths are an important part of many games. Being killed adds
a note of danger, and makes the player a bit more cautious. It adds
atmosphere. LucasArts did an excellent job in eliminating player deaths
from The Secret of Monkey Island, but that game has a light-hearted tone.
The player isn't supposed to feel fear. Can anyone imagine, however,
The Lurking Horror without death?

Given that the game, then, will be placed into an unsolvable state on
occasion, the question is when this is a good idea.

Personally, I think that nothing is more frustrating than to perform a
perfectly logical action, and discover many turns in the future that,
by doing so, you have made the game unsolvable. (This is, in face,
Graham's fifth rule in his `Bill of Player Rights': ``Not to have the
game closed off without warning.''

Your example of reading the poetry before opening the coal bunker is a
perfect case of this. There is no warning that the game has been
closed off, and a player may well spend many turns trying to open the
bunker from the outside. (I would fix this, incidentally, by moving the
wrench to the tool shed.)

Despite the above, however, I still feel that there are times when the
game may become closed off without killing the player. Your case of
time-limited puzzles is, again, a good point. In Zork I, for example,
if the player accidentally breaks the pipes in the dam control room,
the game will become unsolvable if the water is allowed to fill the
room while some essential action in there remains unfinished. I don't
view this as unfair -- any player is going to realize that flooding the
room was a bad idea.

I would say that a reasonable test to see if closing the game off is fair
at any one point is to ask three questions:

o After having rendered the game unsolvable, how likely is the player
to save the game, writing over an old save?
o How far will the player be able to proceed before running into
the newly-opposed barrier? (For example, if the player gives the
jeweled key to the troll instead of the banana, how long can the
player continue before being blocked by the jeweled door?)
o If the player closes off the game and then saves over all of her
old save files, how much work will she need to redo to return the
game to the previous state?

Ideally, the answers to the three questions should be; not very, a short
distance, and very little.

The best way to make an event which closes the game off reasonable is,
in my opinion, to make it fairly obvious that something undesirable has
happened. If the player burns down Manhattan, or crushes a small child
under an avalanche, it should come as no surprise that the game may no
longer be won.

So, overall, it comes down to the designer's preference. Use common sense.
Ask yourself if you, as a player, would find the puzzle fair. Listen to
your playtesters. And remember that if players spend hundreds of turns
looking for the solution to an insolvable problem, or are forced to play
the game from the beginning due to a mistake made well into the game,
you will be limiting your audience.

- Damien

Andrew C. Plotkin

unread,
Nov 27, 1994, 2:58:08 PM11/27/94
to
In article <3ba7a6$6...@lyra.csx.cam.ac.uk>,
Gareth Rees <gd...@cl.cam.ac.uk> wrote:
>
>I'd like to canvas the opinions of the people who post to this newsgroup
>about the pros and cons of letting games get into states from which they
>can't be solved.

I've posted my opinions before, so I'll just summarize: To eliminate
such states (ie, making a completely unlosable game) eliminates many of
the kinds of puzzles I like best. One-use objects are the canonical
example, when several of them fit into puzzles that that each have
several possible partial solutions.

I dislike the sort of game where everything has exactly one use, and
when you use it you know you don't have to consider it in later puzzles.
The obvious way to fix this is the sort of puzzle I'm talking about; you
can get past the X using the Y, and do some more of the game, but later
you realize you need the Y for Z, so you have to go back and think about
everything you've done, looking for a different way past X.

> Personally, I think that nothing is more frustrating than to perform a
> perfectly logical action, and discover many turns in the future that,
> by doing so, you have made the game unsolvable. (This is, in face,
> Graham's fifth rule in his `Bill of Player Rights': ``Not to have the
> game closed off without warning.''

Whereas I think it's a reasonable part of the game; the player can do
many things at that point in the game, but doing them in some orders is
better than other orders. He will have to try several.

> I would say that a reasonable test to see if closing the game off is fair
> at any one point is to ask three questions:

> o After having rendered the game unsolvable, how likely is the player
> to save the game, writing over an old save?
> o How far will the player be able to proceed before running into
> the newly-opposed barrier? (For example, if the player gives the
> jeweled key to the troll instead of the banana, how long can the
> player continue before being blocked by the jeweled door?)
> o If the player closes off the game and then saves over all of her
> old save files, how much work will she need to redo to return the
> game to the previous state?

These are relevant things to consider, but I don't think that they
define a fair game (or section of a game.) Maybe a hard or advanced game
(which, IMHO, Curses definitely is.) An experienced player never
overwrites his old save files, exactly because he is expecting this sort
of thing.

But then we're back to the question of whether a game should be written
for experienced players, novices, or can it be written for both. Which I
don't want to get into. I can only say that I have had a whole lot of
fun doing the 2/3 of Curses that I've completed, and exploring the
ordering-of-subtasks problem has been a large part of that fun. And I
have 17 save files.

--Z

"And Aholibamah bare Jeush, and Jaalam, and Korah: these were the borogoves..."

Greg Ewing

unread,
Nov 27, 1994, 5:03:14 PM11/27/94
to

In article <4iqCHE200...@andrew.cmu.edu>, "Andrew C. Plotkin" <ap...@andrew.cmu.edu> writes:
|> An experienced player never
|> overwrites his old save files, exactly because he is expecting this sort
|> of thing.

Since this is clearly the safest method of save file
management, perhaps support for it should be built-in.
Append a version number to each save file name, and
increment it with each save, unless the player explicitly
requests otherwise.

Greg Ewing, Computer Science Dept, +--------------------------------------+
University of Canterbury, | A citizen of NewZealandCorp, a |
Christchurch, New Zealand | wholly-owned subsidiary of Japan Inc.|
gr...@cosc.canterbury.ac.nz +--------------------------------------+

Andrew C. Plotkin

unread,
Nov 28, 1994, 1:07:04 PM11/28/94
to
I have meditated further upon this subject... (well, a little.) I
previously objected to the Gareth Rees's criteria; let me explain in
more detail why...

> I would say that a reasonable test to see if closing the game off is fair
> at any one point is to ask three questions:

> o After having rendered the game unsolvable, how likely is the player
> to save the game, writing over an old save?

Agreed -- but I prefer to give the player credit for recognizing
situations where it is stupid to overwrite an old save file. Teleporting
somewhere is dangerous. Using up a one-use item is dangerous. If you're
at a decision point between several subplots, and you don't know which
to enter, entering any of them may be dangerous (so save). Etc.

This brings up a separate point which you might want to add: "How much
warning does a player have that he is about to render the game
unsolvable?" Obviously if you see a slippery slide, you know that
jumping down it might be dangerous. The situation in Curses that you
were complaining about may violate this one; you don't expect that
reading a book might get you stuck. (On the other hand, if you've
already encountered a book that teleports you when you read it, that's
fair warning.) Looking things up on the map and suddenly being stuck is
similarly surprising.

> o How far will the player be able to proceed before running into
> the newly-opposed barrier? (For example, if the player gives the
> jeweled key to the troll instead of the banana, how long can the
> player continue before being blocked by the jeweled door?)

I still maintain that this is not a problem. In fact, I like to allow
people to go *farther* after the critical decision; it's a technique for
letting the player see more of the game, early on, thus keeping his
interest.

> o If the player closes off the game and then saves over all of her
> old save files, how much work will she need to redo to return the
> game to the previous state?

I've found that in most text games, this is very little effort. It's
surprising how quickly you can go through the early stages of Curses.
(I've had to start over a couple of times -- not due to losing old save
files; it was because of new versions of the game, and starting play on
new platforms.)

It is characteristic of text games that once you know the solution to a
puzzle, you can redo it immediately (in just a couple of commands); this
characteristic has influenced the development of game-plotting as we
know it. This was brought home to me when I started considering writing
a non-text game with graphic manipulation puzzles, but which was
losable. Since that sort of puzzle can be a major pain to solve, even
after you've solved it once, it would be very unfair to say "sorry, you
used the Magic Hoople in the wrong place, you'll have to restore to an
earlier point and go through this branch again."

Mathematical Institute, (0865) 2-73525

unread,
Nov 28, 1994, 2:11:48 PM11/28/94
to
In article <3ba7a6$6...@lyra.csx.cam.ac.uk>, gd...@cl.cam.ac.uk (Gareth Rees) writes:
> Some minor spoilers for "Curses" in the following:
>
> I'd like to canvas the opinions of the people who post to this newsgroup
> about the pros and cons of letting games get into states from which they
> can't be solved.
>
> "Curses" seems to be a well-respected game, but it's ludicrously easy to
> get the game into an unsolvable state without realising it (e.g. by
> reading the poetry before getting the attic key, or visiting the Tarot
> cards in the wrong order).
>

If I could design Curses all over again, I'd do a better job of it. That
said, there are only about three or four nasty ways you can really lose
at Curses, and I did go to some effort to keep it to so few!

The trouble is that to some extent the alternative can be very contrived.
Setting a bomb off, for instance, is a career decision: you don't get the
bomb back, and you don't get the floor back either.

Graham Nelson

Greg Ewing

unread,
Nov 28, 1994, 8:35:01 PM11/28/94
to

In article <1994Nov28.191148.27767@oxvaxd>, nel...@vax.oxford.ac.uk (Mathematical Institute, (0865) 2-73525) writes:
|> Setting a bomb off, for instance, is a career decision: you don't get the
|> bomb back, and you don't get the floor back either.

#ifdef politically_correct
#define he he_or_she
#endif

I don't think this is a problem, since it's pretty obvious
that setting off a bomb is going to be an irreversible
operation. If the player has any sense he'll save at
that point, and make sure he keeps the save file until
he's sure he's made the right decision!

|> Graham Nelson

Mike Roberts

unread,
Nov 29, 1994, 3:00:44 AM11/29/94
to
Gareth Rees <gd...@cl.cam.ac.uk> writes:

> getting the game into an unsolvable state is running the risk of causing
> the player to lose interest, and should be avoided as much as possible.

The LucasArts adventure games totally sold me on this idea. I've
been slowly forming some idea of how to write good adventure games,
and one of my central themes is that the game shouldn't be tedious.
If you use this as a rule, you arrive at all sorts of useful
conclusions: Don't build gigantic mazes (sorry about Deep Space
Drifter; I hadn't figured things out quite as well when we were
writing that game). Once the player figures out a puzzle, don't
make them keep solving it (for example, once a door is opened, don't
make the player keep entering the 128-digit combination each time
they want to go back through the door -- just leave it open).
Don't strand the player in an unwinnable situation: once you're in
an unwinnable position, you must start the game over (or go back to
a saved position), which is tedious.

If you haven't tried any of the LucasArts games, give one a try --
Sam & Max and Day of the Tentacle are both a lot of fun. They promise
never to kill or strand you, but are fun and satisfying games anyway.
The knowledge that you'll never find yourself an unwinnable position
makes the games incredibly (if you haven't played them) pleasant --
you just know not to worry about capricious game designers, so you can
play the game in peace without constantly saving.

> If you try to avoid this possibility then there are big restrictions on
> the kind of puzzles you can design. No puzzles that have time limits -


> NPCs can't get bored and wander away - opportunities have to keep

> re-presenting themselves...

It definitely requires some extra work for the game designer. If you
start off with the goal of eliminating unwinnable positions, though,
it's really not that hard most of the time. The point about
opportunities is really the key -- you can have timed puzzles, bored
NPC's, and so on, as long as a missed opportunity isn't forever lost:
you have to come up with some excuse for the NPC to regain interest,
or for the timed sequence to happen again. You also have to find a
way to regenerate lost or destroyed objects -- the easiest way is to
avoid destroying anything in the first place, but if you really must,
you can always put the original back where it came from.

*New thread suggestion*: Let's see who can stump whom. Someone propose
a scenario involving an unwinnable situation, and the rest of us (or I,
anyway, if no one else is game) will try to find a way to keep it
winnable. Maybe we'll be able to develop a taxonomy of unwinnable
conditions and the corresponding "fixes" that can serve as a guide for
game authors who want to make their games unlosable.

--
Mike Roberts mrob...@hinrg.starconn.com
High Energy Software 415 493 2430 (Voice)
PO Box 50422, Palo Alto, CA 94303 415 493 2420 (BBS)

Carl de Marcken

unread,
Nov 29, 1994, 1:27:32 PM11/29/94
to
> *New thread suggestion*: Let's see who can stump whom. Someone propose
> a scenario involving an unwinnable situation, and the rest of us (or I,
> anyway, if no one else is game) will try to find a way to keep it
> winnable. Maybe we'll be able to develop a taxonomy of unwinnable
> conditions and the corresponding "fixes" that can serve as a guide for
> game authors who want to make their games unlosable.

It is no problem, I don't think, to restrict a players actions to the
point that a game remains solvable. But it's a different matter
to do this in such a way that you are not giving a puzzle away.

GC SPOILERS
-----------

In GC, there is some Acme whiteboard cleaner in a spritzer- toxic, nasty Acme
whiteboard cleaner. Most players pour it on the ground immediately after
seeing its effects, and it disappears.

I made a mistake, by any reasonable game code of ethics, by requiring
that the player use the cleaner later in the game. MUCH LATER, so
that it was a very difficult puzzle even to remember that the whiteboard
cleaner existed. This was a bad puzzle, though GC was meant to be very
hard and nobody complained.

It would have been easy to make it impossible to permanently get rid of the
cleaner, by only allowing liquids to pass from container to container, for
instance (and eliminating the /dev/null). But this would have been STRANGE,
and any player that knew the game would never get into an unsolvable state
would wonder whether the reason he couldn't get rid of the cleaner was because
it was needed later. This would give the puzzle away.

Carl


Carl Muckenhoupt

unread,
Nov 30, 1994, 8:13:09 AM11/30/94
to
The whole idea of games where it's never impossible to win owes its
popularity, to a large degree, to Lucasfilm Games, who have been making
excellent games on that model for several years. But they have a greater
imperative to do so than us. They make graphic adventures, where playing
a scene twice means sitting through the same animation twice. Cut-scenes
grow tedious through repetition. Even the time spent walking across the
screen becomes a bore.
In text adventures, we have none of these disadvantages. Replaying a
text game from the beginning is fast. Rather than sit through animation,
you just keep half an eye on the output as it skims by, to make sure it
doesn't contain error messages. Game state isn't such a precious thing
when it can be reproduced with so little effort, and, being of small
value, it is only a small tragedy when it is ruined. Small tragedies are
an essential part of games, delivering the sense of challenge.

I speak as one who recently had to play GC from the beginning when I
realized the reason the cat wouldn't come out of the box. This
realization carried the same sense of satisfaction as solving a puzzle,
and as such was worth the minor inconvenience it caused.


--
Carl Muckenhoupt
Breakfast Dragon
-==(UDIC)==-

Gareth Rees

unread,
Dec 2, 1994, 10:22:40 AM12/2/94
to
Mike Roberts (mrob...@hinrg.starconn.com) writes:
> *New thread suggestion*: Let's see who can stump whom. Someone propose
> a scenario involving an unwinnable situation, and the rest of us (or I,
> anyway, if no one else is game) will try to find a way to keep it
> winnable. Maybe we'll be able to develop a taxonomy of unwinnable
> conditions and the corresponding "fixes" that can serve as a guide for
> game authors who want to make their games unlosable.

Suggestion (pace "Border Zone"). You're trying to smuggle in contraband
items (drugs, weapons, samizdat) into a totalitarian country. At the
border, guards search you, discover the contraband and confiscate it.

--
Gareth Rees

Inigo

unread,
Dec 2, 1994, 10:45:13 AM12/2/94
to
Gareth Rees (gd...@cl.cam.ac.uk) wrote:
: Suggestion (pace "Border Zone"). You're trying to smuggle in contraband

: items (drugs, weapons, samizdat) into a totalitarian country. At the
: border, guards search you, discover the contraband and confiscate it.

Obvious solution: You catch the next train back, get some more
contraband, and try again.

Another solution: Steal the contraband back from the guards.

Another solution: Maybe you're ultimately trying to bring the government
down. Failing to smuggle in the contraband is now impossible, but there
may be other paths to reach the final goal.

The problem isn't that a situation is inherently unwinnable (unless you've
just blown up the world or something), but the effort needed on the
part of the programmer to keep the game winnable just isn't worth it.
It would be better used improving the game better elsewhere, and it isn't
really much hardship for the player to go back to a savefile, or even to
start again.

--
Inigo Surguy http://www.csv.warwick.ac.uk/~phunc/

Mathematical Institute, (0865) 2-73525

unread,
Dec 2, 1994, 12:00:18 PM12/2/94
to
In article <3bne40$i...@lyra.csx.cam.ac.uk>, gd...@cl.cam.ac.uk (Gareth Rees) writes:
> Mike Roberts (mrob...@hinrg.starconn.com) writes:
>> *New thread suggestion*: Let's see who can stump whom. Someone propose
>> a scenario involving an unwinnable situation, and the rest of us (or I,
>> anyway, if no one else is game) will try to find a way to keep it
>> winnable. Maybe we'll be able to develop a taxonomy of unwinnable
>> conditions and the corresponding "fixes" that can serve as a guide for
>> game authors who want to make their games unlosable.
>

Suggestion (pace Nigel Kneale's late 1970s "Quatermass" TV play).

You discover that the only way to stop the aliens from killing the Earth's
population with a giant harvesting beam is to set off an atomic bomb right
in the path of the beam (which, unfortunately, will destroy the entire
county the game is set in). You're dead either way - but is Earth saved?

Incidentally, I have once played a completely unlosable graphical adventure,
a peculiar but wonderfully well designed game for a BBC Micro called, um,
- can't remember. Odd one-word title. Anyway it was a platforms and ladders
style game but with adventurish problems - bows and arrows, levers, that sort
of thing. What was clever was that you can always, always rescue the
situation, fetch a fallen missed arrow, etc.

Graham Nelson
Oxford, UK

Reply all
Reply to author
Forward
0 new messages