There are two ideal types of text adventure games. Between them is a continuous
spectrum where most, if not all, existing games can be placed. No actual game
fulfils all the requirements of neither one of the ideal types, but some can be
placed further out on the left or right flank of the spectrum than others.
I would like to call the two types simulationist-oriented and decision-oriented.
One of the differences between them is that movement is predominantly spacial in
the simulationist-oriented type, while in the decision-oriented type movement is
In a simulationist-oriented game, like Anchorhead, the player is presented with
a large, open and easily accessible game world where the PC can move around,
pretty much, unhindered. Movement in time, in a simulationist-oriented game, is
limited and PC-dependent. By "PC-dependent" I mean that time only moves when the
PC triggers the movement by performing certain actions (Leisure Suit Larry). By
"limited" I mean that time doesn't move continuously and symmetrically (one
action = one minute) but sporadically and unrelated to the number of actions the
PC has performed. When time moves in a simulationist-oriented game it often does
so with a big leap. The game state is changed in an important and pervasive way
(Wishbringer). Between the leaps time is frozen. No matter how long the PC
strolls and pokes around in Anchorhead, the fishermen in the pub are still going
to drink their beer and look gloomy and the rain is still going to fall. Unless,
of course, the PC manages to perform the action that triggers movement in time.
Where there's no time there's no memory. That's why the PC can ask an NPC, in a
simulationist-oriented game, the same question ten times in a row without
eliciting any irritation.
In a decision-oriented game spacial movement is severely restricted. In the
opening scene of Photopia the PC is speeding away in a car, but the car is just
a piece of props and there's no freedom of movement. Instead there's a
continuous movement in time created by the conversation with the driver.
The actions available to the player, in a decision-oriented game, are usually
limited in number, non-trivial and irreversable. These are perhaps the most
important distinctions between the decision-oriented type and the
simulationist-oriented type. The possible actions the player can choose from in
Photopia or Galatea or much fewer than in Anchorhead. In return they are
important to the plot and can not be reversed. By opening a door in a
decision-oriented game the player closes other doors. The vast majority of
actions available to the player in a simulationist-oriented game are trivial and
reversable. This is perhaps best exemplified by movement by cardinal directions
and the spacial symmetry it allows.
It would be a mistake to assume that the decision-oriented game is somehow less
"interactive" than the simulationist-oriented game, just because it presents the
player with fewer actions to choose from. One could, just as easily, claim the
opposite by saying that the actions available to the player in a
decision-oriented game have more meaning and weight by being irreversable.
One of the most conspicuous characteristics that distinguishes
simulationist-oriented games from decision-oriented games is the different types
of conversation systems they use. While the simulationist-oriented game uses the
traditional "ask someone about something" approach and treats the NPC as a mere
receptacle of information, the decision-oriented game relys on a complex
menu-based conversation system (Rameses) or an intricate system of flags that
keeps track of the conversations (Galatea, Varicella). The reason why
decision-oriented games rely so heavily on conversation is that conversation is
an elegant and interesting way of creating movement in time. For a similar
reason simulationist-oriented games rely heavily on cardinal directions; it's an
easy and practical way of creating movement in space.
I disagree. There are more than two types, but I don't know what they
are, I'm just betting on the law of averages. Also, it is possible,
as you well know, to use more than one type in a single game.
Welcome to the big purple beanbag, where people tend to think more
Resident, on call.
That's an interesting way to look at it. I think the simulation vs
decision-tree dichotomy is useful, but I'm not sure it's truly the case that
simulation implies pure spatial orientation or that a decision-tree style
implies pure temporal orientation. As counter-examples to the former,
consider the Infocom mystery games - clearly simulationist - where the clock
ticks away, and the plot proceeds, regardless of the PC's actions.
It seems to me that the way the clock works in Anchorhead is actually a
departure from simulationism. In fact, the points at which the clock moves
fit your criteria for decision-tree games: these changes are irreversible,
and the range of possible outcomes at these points is strictly limited. I
think it's arguable, then, that Anchorhead is not at an extreme of
simulationism, but rather a hybrid that has simulationist-style freedom of
movement and action within temporal sections gated by decision-tree style
plot development. Personally, I do think this is a pretty ideal design, but
I don't think it's at an extreme of simulationism.
> Where there's no time there's no memory. That's why the
> PC can ask an NPC, in a simulationist-oriented game, the
> same question ten times in a row without eliciting any
I don't think the degree of conversation "memory" is necessarily a
distinguishing feature of simulationism vs decision-tree. As another
example, consider the LucasArts adventures: they're pretty standard
simulationist IF (freedom of movement around numerous locations, many
manipulable objects and large inventories, a fair number of verbs), but they
use a menu-tree conversation style where NPC's do have memories, not only of
the position within a specific conversation but also of previous
conversations, and sometimes even of events outside of conversations.
Anchorhead itself is a counterexample in places, too: pester Michael with
too many questions at one point and he bolts; grill another character about
a particular subject enough and he finally cracks and reveals a secret.
My feeling is that your observation about conversation memory is not exactly
inaccurate, but that this is something of a coincidence rather than a
defining feature of simulationist games. I'll propose a couple of possible
reasons for the coincidence.
First, in many cases, it might just be a matter of resources: programming a
big simulation might just take so much effort that it doesn't leave a lot of
bandwidth for developing a deep conversation system as well. The author
doesn't so much choose not to give NPC's memory and so forth, but rather
chooses to spend his or her time on other things.
Second, I think the usual "statelessness" of conversation in simulationist
games is really a result of the tendency of text-based simulationist works
to use the ASK/TELL format, and I'd further claim that the reason they use
ASK/TELL in the first place comes from user interface concerns, not from
anything inherent in the simulationism. Graphical simulationist IF (such as
the LucasArts games) does not tend to use ASK/TELL, and as I claimed above,
their conversations tend to be more stateful than in text games. Text IF
uses ASK/TELL so frequently, I think, because it's the most seamless way of
integrating conversation into a UI where everything else is done by typing
in commands and everything else about the UI is modeless. Seamlessness is
especially important in simulationist games because the immersion in these
has to come from the game mechanics, so the UI has to be as invisible as
possible; in a decision-tree game, the emphasis is less on the game
mechanics, so a more visible UI is often acceptable. Given this ASK/TELL
bias, stateful conversations become difficult to code (though not
impossible, as Galatea demonstrates), both because of the technical
complexity and because of the playability implications.