Stating intentions

I have been thinking about the conflict between player freedom and having a coherent story. One example is foreshadowing: the author might carefully place hints about an upcoming plot point, but if the player is free to bypass it then the story might not make as much sense.

One possibility is to let the player somehow state his intentions during the game. The story can then unfold in an appropriate way, and the game will have a clearer understanding of the player’s actions (and so give more relevant responses).

This (kind of) solves the “coherent story” problem: if the player commits to a certain direction, he is likely to stick to it – and if he doesn’t then he can’t complain about the story not being coherent, since he’s the one who broke it.

Intentions are simplest to express in a choice based game:

It seems the final piece of the Great Jewel is going to be first prize in the deadly, dastardly marathon of doom. Do you want to (1) train to enter the race, (2) steal the jewel, or (3) bribe an official to swap it for a replica?

In a parser based game, there could be times when the player is asked to make an explicit choice like the above example (perhaps with the option of changing intentions later on), or there could be a command to express an intention: “I want to …” – which could be tricky to do, since there are so many possibilities. A compromise might be to have a menu (or command) that lists all currently possible intentions.

For the above example, the game could work just fine without needing to know the player’s intention. But if their intention was known, this could change the way the story is written: training for the games might take weeks, with its own interesting sub-plots; obtaining a replica could require a trip to another city; stealing the jewel could require planning, observing the guards and so on. Even if the choice doesn’t result in three separate sub-plots, the location descriptions and responses to PC actions could still be customised based on what the game knows the player is trying to do.

So: has this been done before in parser IF?

(Afterthought: all players really do is enter their intention every turn, e.g. “go north” – just on a much smaller scale than I am talking about here).

You can easily envision a game where, for example, there’s a fire axe in a glass case on the wall at one point – and the player can either bypass it, or break the case and take it (indicating their intention to commit murder later in the game).

Indeed, this kind of interaction can operate as foreshadowing even without explicit game logic.

In the Erudition Chamber, you are told upfront that you will be sorted into one of four groups depending on whether you approach challenges with brawn, brains, tools, etc. To get the best endings, you need to plan ahead of time which route you will go down.

But the game is rather artificial, and you never have to declare your intention.

In Tapestry, you have three challenges. In the first challenge, you choose one of three paths (confrontation, fleeing, or acceptance), and this restricts how you can solve future puzzles.

On a side note, I loved the Lucasarts Indiana Jones game about Atlantis where you choose to proceed with puzzles, combat, or teamwork. But it was a graphical adventure; I only brought it up because I loved it.

If you clearly telegraph consequences to the player, then you can derive the player’s intentions based on their choice of consequences. Or, to put it another way: the best way to figure out what the player wants is to ask the player.

This can be explicit. Example: “Do you want to steal the emeralds to appease your former masters, or seek peace with your victims and ally against your former masters?”

But it can also be relatively subtle. Example: “A key turns in the lock. There’s only a moment to decide whether you will reveal yourself to the tirewoman, thus abandoning your hope of stealing the emeralds, or conceal yourself and hope for a second chance later.”

This second example could easily be choice-based:

  1. Wait for the door to open and reveal myself.
  2. Hide in the closet.

…but it could just as easily be parser-based. Any attempt to conceal yourself takes path #2, and any command that fails to do so takes path #1.

This is a standard approach in Choice of Games, where potential consequences are often explicitly described before the player makes a choice. I can’t think of any parser examples offhand, though.

I agree that there are lots of ways (and it’s probably preferable) for the game to work out the player’s intentions based on their actions, but I still like the idea of letting the player be explicit about it sometimes, since it could customise the way the game responds – and allow more than the usual amount of player freedom. An extreme example is a sandboxy game like the one in this thread, which could trigger events related to the player’s current goals (and mention things relevant to those goals in descriptions).

I came across this paper with an interesting idea: generate a hypothesis about the player’s intentions and then test it. This is in the context of story generation, though, not normal IF:

Player-Protagonist Motivation in First-Person Interactive Drama (pdf)

Not necessarily. There are many uses for an axe, and in most games taking everything that isn’t nailed down is standard behaviour anyway - and if the game wants to dissuade me from indulging in that behaviour it can, for instance, have NPCs react to me holding the axe.

I really dislike it when games sneakily track variables like that, pretending to flesh out my PC as a result of my actions because they think they are reading my mind and know what my actions meant. It’s akin to the “parser that pretends to know what you’re typing and instead is pattern-matching”.

So as far as this topic goes, in my experience, I’d prefer to be able to state my intentions explicitly - and possibly even be given a chance to change my mind at certain points (because life’s like that), and accept that I may be at a disadvantage because of that. The alternative, where the game guesses, is beautiful when it works (i.e., when it’s kept simple, obvious, and/or binary), and SO frustrating when it doesn’t!

Does anyone else remember the Deus Ex bit where you could accidently cause a bloodbath by trying to go through a building’s main door to complete your objective? And this would happen even if you already HAD completed all or most of your objective before and were now just exploring a bit? The game would proceed as though you’d CHOSEN to trigger that bloodbath and damn the consequences. Not cool.

That seems like a problem with failure to telegraph, along the lines of Carolyn’s comment, rather than a problem with action-as-intention per se.

If the game fails to telegraph, then action-as-intention will always be a problem, so they’re pretty closely related. If the game telegraphs well, action-as-intention will almost always be read correctly. You can also do retroactive telegraphing - you pick up the axe thinking it’s going to be a tool and the game goes “It weighs reassuringly in your hands. It feels good to be armed”, you know the game actually probably intends it to be a weapon. If you’re in a zombie survival game, you’re going to keep hold of it, probably; if you’re playing a high-school kid and it suddenly looks like you may be starting down a “high school massacre” where you’re the perpetrator, you may decide to put the axe back immediately, no harm done.

Action-as-intention is only a problem when the game keeps stum about it, which they often do because they try to hide their workings, try to hide they’re games to make “immersive experiences”. So if they get them wrong, and don’t telegraph it before OR after, things go bad. So yeah, I’d say they’re related.

EDIT - Anyway,

I can’t say I agree. If the player skips a plot point, the game shouldn’t continue as though the plot point HAD happened - though it can continue as though it had happened and the payer failed to witness it. Deadline, Varicella, Nightfall and their like are all based on this concept (Nightfall being a bit less extreme). But the story is never broken, and it should never be. If the player has the possibility to break the story, that’s the game’s fault. The player may have the possibility of failing to guide the story to a successful conclusion, or may miss all the crucial events that would make them understand what’s happening - but the story is never “broken”.

In The Mind Of The Master does this.

I meant, if the player states a certain intention explicitly and then doesn’t act consistently with that intention. Maybe “broken” is too strong a word – just that the story won’t be as good as it would have been (there may be unresolved loose ends).

Do you remember any specific examples?


I completely misunderstood you there, then. Sorry! But then it gets easier for the game to police the player, doesn’t it? If the player stated the intention to uphold the law and then starts pickpocketing, the game will be able to come up with appropriate consequences. Like alignment changes in roguelikes like ADOM.

Sure, although you probably are better off just playing the game. It’s from IFComp 2007, so it is short.

You choose a disguise at the beginning, which affects your story path. It feels very gamebook-y in structure with different scenes split up like you suggest.

The second of the Heroes trilogy from Choice of Games tries to do this, and it is super annoying. You develop a character type you’re supposed to stick to, but it makes it so a “full points” play-through of the game so narrowly focused you have to guess at what is essentially a linear route.

I was thinking about this issue today, and while I primarily use twine as a creator, I play a number of different IFs and games where the player can assert their intentions, and to me it’s always a toss up. If you do your job and do it well then I think intentions will be clear, and some can be assumed to be automatic. As someone previously mentioned if your fighting zombies the axe will probably be a weapon, so context helps a lot. Yet there are times when you have decide whether you want players to have information or whether you aren’t being clear. There have been games where you can assert an intention, and I have found myself doing so without information, and it works great. There have been games where an object was limited or an action produced a very specific result, and that was clear. But on the flipside I played one game where I was given a pipe and because of the structure and the objects intention there was a disconnect from the narrative. “You” can’t use it as a weapon, but you encounter someone you need to get past, but unless you went a magical route you had to get another weapon. That destroys immersion for me. It is a sensation I get a lot in horror games and is why I don’t play many of them. Truthfully it may be why I don’t tend to play parser IF often because I do get cognitive dissonance a lot.

There is nothing more frustrating to me as a reader or player than situations where actions produce completely unexpected results. An example may be Bioware games(and apparently they have authors use Twine sometimes). A number of times in recent games I have clicked a dialogue choice and been very disappointed with the actual dialogue and it’s results. If the text says “You’ll be a little mad” and the next section of the story is you punching someone in the face then your intentions aren’t clear. The problem is to someone an intention will be evident, but to a player without proper telegraphing it’s a shock. So you have to be very clever in sneaking your intentions into a scene. Overstating them doesn’t make for a fun read, but they need to be organic, and you need to guide your player towards understanding your intention and choice.