Because I dislike envisioning the protagonist of the story as a monosyllabic verbal cripple who can barely get out a coherent thought and can’t express anything that can’t be encapsulated in one or two words?

Because trying to control a conversation in IF line-by-line is like trying to masturbate wearing boxing gloves.

Because the IF interface manages to avoid modalities in all things except in conversation, where special TALK modes and TOPIC modes abound, modes where I can type the same exact thing as when I am not in that mode, and get entirely different results. This is a sign that the interface simply isn’t suited for the goal and has to be cloned-and-grafted to fit unnatural territory. Watching some parser games dissolve into numbered CYOA-style whenever conversation happens is another very strong sign that a bad graft has occurred.

Lastly and most importantly it’s because I don’t even need to choose words for the conversation in order to actually have and influence a conversation. I am barely kicking in any words anyway so what’s the point? A conversation is just a series of events that happen, nothing more, and there’s no reason it has to be more, so why go there into crap-interface territory? It just isn’t necessary. My character can be afraid and angry and curious and I never have to type ‘be afraid’ or ‘get angry’ or worse, ‘think about curiosity’ – why should conversation be any different? It should just happen.

A verb-object parser is just the wrong tool for this; in fact, since it is fundamentally predicated (NPI) on verb-object, IF in general is simply not well suited for simulating participatory dialogue – full-stop. This doesn’t mean it can’t depict conversations and even have their flow affected by the player: one thing has nothing to do with the other. I can’t decide which way to point my feet in IF, that would be tedious – but that doesn’t mean I can’t affect where I walk. A poor foot-simulator is not necessarily a poor walking-simulator. We just have to get less literal-minded about it to make it graceful.

I have never seen an actual graceful conversation system in IF that I can recall – I’ve seen some heroic attempts, but nothing ever nailed it. So I believe that we have to get way less literal-minded about adding interactivity to conversation, before we can get to the point where I don’t groan inwardly every single time a game takes me into conversation mode. It’s just a huge black mark on a game for me the moment it asks me to supply actual words into a conversation. I get bored out of my skull every time and I don’t mind admitting it – it’s not like I am an action movie guy either or have no patience for slice-of-life type stuff. Quite the opposite, so that’s not it, it’s just about bad interface. The fun part of being in a conversation is not the choosing of topic headings; the real fun is choosing your attitude and word order, so that isn’t captured at all in these systems, it’s just removed, leaving conversations a pale shadow of the liveliness they normally entail, and so conversations would be much better off without these ‘tools’. So the prevailing convention for this just doesn’t work and no tinkering it will fix it, the whole ASK/TELL ABOUT paradigm needs to be discarded completely, IMO, to be replaced with… nothing! Just a general philosophy that the graceful way to influence conversations in IF, is indirectly, or not at all.

This has been a solution in search of a problem, the entire time.

So, ‘Show don’t tell’, applied as a parser principle, to put it rather reductively. Anyone else feel the same about this? Does anyone else have that feeling of disappointment and trepidation every time any IF game wants me to tell it what specifically to talk about and what to ask about?


I haven’t really felt that way before, but I can certainly see what you’re saying.

Can you give a sample transcript of how you envision conversation working? I’m having a hard time seeing an immediately obvious elegant solution.


Well not really because there are no specific commands I’m thinking of, but here’s how it plays out in the one I’m currently writing. You just do things and conversations happen. The player does plenty of things that can spark conversation – you would just key conversational results off things the player picks up or examines or places the character goes and whether people follow, rather than asking the player to specifically do something conversational. Ever notice how when you are talking to someone your physical actions while talking tend to become topics of conversation? It’s pretty simple, it’s – how would you do conversation if there is no conversation system? That to me is still the most elegant system, so that doesn’t speak very well for the systems that are out there.

Why don’t we need a system for deciding what the player feels or thinks to himself? We don’t have one and yet authors have had no trouble at all having the player think or feel things to themselves based on the circumstances – I see no reason why conversation is a different scenario. Dialogue is the same essential scenario as an internal monologue; it just has two thinkers/feelers instead of one – that’s the only difference. So, if internal monologue doesn’t require a whole special set of commands to be portrayed in IF, then why does everyone think it’s so necessary to have special commands for dialogue? To me it’s just an accident of text adventure history that dialogues require special commands to navigate whereas internal monologues just happen based on the situation, because I don’t see a logical reason for distinguishing them and the results of handling dialogues differently from monologues have been pretty poor.


So if the player wants to know about another character’s opinion of x and x isn’t a physical object, do they ‘examine x’?

I’m not sure about the historical accuracy but I can well imagine that conversation in IF originated as mainly another kind of puzzle (figuring out what you have to ask that NPC about) – which may (perhaps) explain why some features of conversation in IF gets in the way when it doesn’t serve the purpose of a puzzle.

Would you say conversation is less bothersome when clearly used as a puzzle?

Perhaps a mere TALK TO command that the player can use to set off conversation, would be less troubling? Anyway, that seems to me to be an action of the same order and in the same mode as ordinary commands like LOOK AT or GIVE TO etc.

Laroquod is right in that IF games could do a whole lot more with the most basic “TALK TO” conversation implementation, based on context and NPC responsiveness to the player’s actions. In some games, this could easily satisfy all the need for NPC-interaction and dialogue. However, I don’t think NPC responsiveness and context-sentive dialogue based on the simple TALK TO command could completely replace ASK/TELL.

The other side of the coin is that the player might feel frustrated when he or she wants to command the PC to interact with an NPC in a specific way, but the NPC doesn’t pick up on any of the PC’s attempts as conversation topics. Sometimes, the player may actually want the PC simply to say a specific and simple question or statement, or to give a one-word answer. In an IF scenario where an NPC asks the PC to find a key, and the player wants to know more about this key, it would probably be less frustrating to have to type “ASK (NPC) ABOUT KEY” than to try to think of a non-conversation command that might make the NPC talk more about it.

I think that responsive NPCs and perhaps a simple “TALK TO” command that is sensitive to context can fulfill the need for more in-depth dialogue than the parser system can naturally accomodate, but I don’t think we’ll ever stop needing ASK/TELL, or other sysytems that let the player directly choose conversation topics (such as CYOA conversation trees).

A handful of games do.

Short answer: agency.

Longer answer: things that the protagonist says aloud to another character are more likely to change the state of the world. There are relatively few games where what you are thinking changes what happens in the story. There are a lot of games where you have to convince an NPC of something, or find out something from him, or establish a relationship, or whatever. It’s possible to have the game work out for you what you should say at every juncture, but that rules out a lot of kinds of gameplay. And if the player is responsible for directing the conversation, he needs some non-accidental way to steer it.

(Also, I realize that you may have been speaking hyperbolically, and certainly the features you describe occur in most current IF that deals with conversation. There are some exceptions, though, if you’re interested and haven’t run into them – this old article hasn’t been updated for years, but it does point to some older examples of various conversation systems; I don’t have a list assembled of games where the NPC chats with you in reaction to other behavior, but here is a brief one of mine.)

As evidence for the prosecution, this is how I handled conversation between the title characters in Walker & Silhouette in most scenes, and a lot of players seemed to think they had a bit of chemistry.

I don’t think any commands are really required in IF. Often the biggest issue I have while developing a game is handling all the default actions I don’t care about. (I finally actually have a WIP where throwing is something other than a nuisance to be disabled.)

But I think it’s important to distinguish between conversation in the sense of two characters talking to one another as flavour, characterisation or divulging essential information, and conversation actions in the sense of things that the player wants their character to say to another character. Certainly, I don’t think anyone has hit on an entirely satisfying way to do this, but I think it’s a healthy impulse in players if you want them to engage with your work as an interactive story.

This, at least, is purely a matter of prose. Yes, if it goes:


Todd says, “You know how you only use use ten percent of your brain? Well, it’s because the other 90 percent is filled up with curds and whey!”[/code]

Then you sound like a Neanderthal, which is why it’s usually better to do this:


You say, “What’d I do? What authority do you represent? You can’t do this! I didn’t do anything! You can’t prove anything! I’M A ROCK STAR!”

The vegan police say, “We have it on record that at 12:27 this afternoon you did knowingly consume a restricted food item. Gelato.”


You say, “Gelato isn’t vegan?”

“It’s milk and eggs, bitch.”[/code]

(Dialogue from Scott Pilgrim book 3, because I’m bad at dialogue.)

Some interesting responses. I’ll run through them…

Yeah. Although, I’m not structuring my puzzles so that getting people to cognitively sniff objects is a regular task. Having someone around versus not having them around is the major decision that’s going to affect the outcome of the game rather than particular things you try to say to them; because if you don’t say it they probably will – which seems to me more like real life. Playing ‘pulling teeth’ with NPCs is just never a fun thing, in my book. I just want them to offer stuff up while I’m thinking about it, which probably means while I’m manipulating it or otherwise trying commands on some other noun that is not the NPC. Then see the conversation is automatically about something important and not just an opportunity to shove a file folder full of backstory at me, wholesale, and expecting me to just page through it for its own sake, absent a narrative. That’s not a fun type of interactivity – why give me the drudge work? It’s not interesting for me to have to control of the order in which I read a glossary – and that’s essentially what the talk model is, treating a person like a dictionary. I mean – these are really boring conversations if you were to consider them in, say, a filmic or a literary light. They’re mostly purely informational though I admit they could be written better by some than by others. But if you want me to read backstory, weave it into the narrative on the way to something else…? Isn’t that generally the wiser way for a writer to win the reader’s indulgence for reams of worldbuilding?

You and Bainespal mentioned TALK TO. I find it marginally better than ask/tell about because it doesn’t implicitly or explicitly) put me in a different command mode. But I don’t really like to have to tell the game when I should talk to people, either. I’m softer on it though – I could be convinced. I’d personally prefer to use the word GREET, though. As for when it’s used as a puzzle. No, my very least favourite puzzle in the world besides huge identical mazes, is the whole figure-out-what-specific-magic-word-to-mention-to-the-NPC thing. It’s like, the NPC is a lock and there is one word that is the key. This has always felt unrealistic and more like a case of my solving a parser than me solving a real life situation.

Yeah, I remembered that afterward. I get an inward double-groan when I see the ‘think about’ command. Usually though it hasn’t been a mandatory part of the puzzle so I’ve been able to just avoid going there.

Yeah, you’re absolutely right about agency, and that’s exactly my objection is to the placement of player agency itself, not just to the ask/tell syntax — I object to the player’s agency being extended to cover individual words in a conversation. I realise that it is possible (now, increasingly common) to construct puzzles that require extending the player’s agency that far — I just don’t prefer those kinds of puzzles. It’s not interesting to me. I would prefer it if the author rearranged things such that I am in charge of something else other than dialogue and can still affect the outcome. I don’t want control of that; for whatever reason, perhaps having to do with my theories as stated above or perhaps some other factor I haven’t figured out yet, it’s just not interesting to me.

I believe I read that article a while back. It’s a good article, I like reading people’s theories about it because I’ve always been curious about why it works so poorly for me. It’s something I’ve been thinking about for a long time, in fact, and I’m dismayed getting back into zcode games how popular conversational commands have really become – they are in so many games, now.

I have heard lots of good stuff about W&S and it’s high on my list to try – forgive my ignorance of some of the experiments that have already gone on. The best example I could think of for conversation being handled in IF is in Violet, where of course it’s all represented as an internal dialogue with an imaginary character – no particular commands required, it just develops with the story. That felt like a real relationship. Relationships are based on conversation of course, and this elaborate conversational syntax we’ve developed is like an albatross over every character’s neck converting them from a person into a reference manual – remove it and it feels like sudden, relationship freedom, at least to me.

BTW I’ve been wanting to congratulate you on Rogue of the Multiverse – really enjoyed that one! 8)

This approaches Emily’s sense of agency I think. Like not just how to depict a conversation but how to affect its flow. I think of the character as separate – if I want to change what my character says then my favourite goal is to change the state of the character, to change ‘my’ attitude or what I’m likely to talk about. Via the things I do.

Thanks again folks, lots of thought-provoking stuff here, I’m glad I blurted out my big secret peeve, lol…


It’s a good point that you can dress it up so that it sounds more natural. However, in terms of player agency in conversations, which is a much clearer way to frame my objections I’m discovering, you are still only adding single words into the interactive mix so your steering cannot possibly be intelligent enough to do justice to the sort of control you should have to make a conversation feel both dynamic/real and interactive. So I guess I’m just saying, why is it so important to even achieve this? We can have great dynamic/real conversations – just because it’s an interactive story doesn’t mean every aspect needs to be under player control. As the writer you choose what requires player agency by how you design your puzzles, I just don’t get what’s so great about getting the player’s command line all entangled in a sort of word-by-word, dialogue-by-inches. What’s the upside for the player? It doesn’t actually feel like you are half of an interesting conversation, so that’s not a win. What is the win, interactively? I don’t see it. I want to give the player things to do that feel real, are fun or otherwise interesting to control, and pay off well. I just don’t see how the prevailing talk conventions will help me fulfill those goals.

I’d say “the sort of control you should have” is control that focuses the player on what matters to this game. Nothing in a game is ever interactive in all the ways real life is interactive; by picking the type of interactivity you want to make available, you’re establishing something about the kind of story that can happen.

Sometimes you want the player to focus on tone (Varicella, Forever Always); sometimes on getting clues and exploring information (lots of games, but to pick a great one: Make It Good). Sometimes it’s about picking from the limited set of things that a well-defined protagonist would do in this situation (Rameses).

I think the kind of interaction you describe is appropriate only for a subset of the kinds of stories I would like to tell. Specifically: it works for a strongly defined protagonist and strongly defined NPCs, within a small, dense setting where many or most actions will carry some kind of emotional resonance, and where there is a single definite goal that the protagonist is working towards. Great for Violet. Great for other one-room games, or games that are highly railroaded through a linear story and sequence of locations (as much of Pacian’s work is).

It’s not so great for mysteries. It gets in the way if you want the protagonist to be more freely defined by the player’s actions. It’s nearly impossible to pull off in a large open world, or in places where you’re giving the player major choices about what to accomplish.

And, I would argue, it rarely works in cases where the protagonist’s main focus and goal is to do something with or to an NPC. You can have games where the protagonist is doing something else (like, say, trying to finish a paper, or rescue his lover from a hanging) where an NPC is strongly motivated to care about what you’re doing or how you’re doing it. But if the whole point of the scene, from the protagonist’s point of view, is to talk to that NPC? It’s going to feel odd, or at least extremely stylized, if you’re only allowed to do it indirectly. “You’re a politician. Convince this group of rich donors to back you. P.S.: You may only achieve this by physically manipulating the contents of the room.”

It’s been done. Textfire Golf is almost that game. But… stylized, like I said. It definitely wouldn’t work for everything.

I spent ages programming every character in Aurora to respond when asked about every other character, so I may have a vested interest. Anyway:

What about the ‘hint’ command? Any game that isn’t utterly self-explanatory needs some way to let the player know what to do next; the trick is making it not break immersion. So why not ask other people what to do next? It beats having a metagame hint command, IMO.

I can think of situations–not even necessarily puzzles–where you might want to control who you do or don’t say something to. What if the NPCs’ motives are unknown and you don’t know who to trust? >TELL IRIS NETWORK ABOUT DOMZ may return a very different result from >TELL ALPHA SECTIONS ABOUT DOMZ.

In that situation, though, you want the third option of actively misleading the people you don’t trust. Has anyone written a game where there’s a command for telling the truth and a command for lying? Because that would be awesome!

It can feel that way if one command triggers a long conversation.

Yes, a single command triggering a long conversation may seem no better than a long conversation triggering automatically, but there’s a key difference: You can choose not to have the conversation. If you’re trying to figure out how to get into the back room of the bar and all the bar patrons keep talking to you, it could get annoying; you could miss important stuff skimming to get back to the puzzle you were working on; it even breaks character, because your character should be focusing on the puzzle too.

I think your points are generally worthwhile, and help me to think about conversation, but I disagree with the point that no game has yet featured believable, non-clunk dialogue with agency. Personally, I find the conversation model used in Galatea and, later, in Blue Lacuna, entirely satisfying – in that it at least takes conversation to the same level of control and immersion, despite occasional absurdity, found in all the interactions in IF. Overlaying keywords and topic suggestions on top of ASK/TELL, if the writing is skilful, is a damn good solution, I think.

Forbidden Castle did an okay job with this- for being 1985.

Basically it assumed that any text that wasn’t a command was a converastion attempt if NPCs were present. (There were 6 or 7 people to talk to).

This worked well, but it could be a pain when you made a typo. You could specify NPCs with [person], [text]. And it wasn’t as good as a real conversation.

As a beginner, it helped me avoid the ASK/TELL/SHOW nonsense and made me feel that the game was listening to me. (It also would allow you to examine ANYTHING, and if the object wasn’t recognised it would say “A quick look is all you needed” or something to make it seem like objects that weren’t implemented were.)

It seems like there may be a case for “If the topic understood includes ‘KING’ and the topic understood includes ‘CASTLE’” modes for conversation, or having the parser assume that non-commands are a case for conversation.

Something along the lines of

Instead of [creating a parser error]:
  if a person is visible (called the guy):
    try saying the player's input to the guy;
    [continue the parser error]

Not sure how the coding on that would go.

As regards entering a different way of operating within IF just for the conversations, I once attempted a different way of tackling ASK/TELL - “, ?” and ", " would ASK about the keyword, and “, .” and “, !” would TELL about it.

My goal was, of course, to make it feel as though it’s part of the same mechanism. You can order people around with this formula, which is in itself a proper and logical extension of your own commands - so why not use it for conversation, as well?

I got it to work very nicely, too. Pity that it made actually ORDERING people around harder to capture, but I wasn’t expecting on using THAT feature.

Personally, I don’t see anything wrong with ASK/TELL… sure, when misused it can be trying and awful, but when misused, everything is awful. ASK/TELL allows the player to communicate an idea in its simplest form. Which is what the parser in IF does - communicate an idea in its simplest form. Not “read out page number one of the book with the red cover sitting on the table.” Just “GET BOOK”, “READ IT”. Similarly, not “Jane, don’t you think Max has been acting strange lately? I’m getting worried about him.” Just “ASK/TELL JANE ABOUT MAX”. Trying to get fancy with that is akin using adverbs - “GO NORTH SLOWLY”. And we all know what a kettle of worms adverbs are, for author AND player.

I’m not saying the system is perfect, I’m just kinda surprised so many people considered it so inadequate. Anchorhead remains to me as an example of how to make ask/tell create serious conversations - I asked about what I wanted, I got told about it, I asked about something else. It was so real!

EDIT - “Kettle of worms”?! Way to mix metaphors. I blame 1am.

I’ve been thinking about this stuff lately, because I was thinking about what kind of next game I’d wanna make. I’m aware that I don’t like conversation in IF either, because it always feels klunky to me, broadly because I have the same feelings as Larroquod.

I note the usual conversation commands didn’t bother me much in old Infocom games, and that’s probably because expectation was lower. I didn’t expect the NPC to talk about anything but a handful of preprogrammed topics. Today if you take that approach, people would probably yell at you for having robotic NPCs. But to me, they’re still kinda robotic, even if they can talk about 20 topics instead of 3.

People are so complicated that if it comes down to the level where I have to specify what to talk about via the parser, I’ll always find 200 holes where the NPC spits out a default blocking response, that makes them feel klunky in a way I never mind when I’m dealing with the physical environment - because the trees and doors don’t have personality, etc.

Obviously part of this is taste, and also what’s best for any particular game’s circumstances.

Blue Lacuna, the conversation there was like the nicest CYOA format you could have for conversation in text, but it still felt to me like CYOA based on specific words.

Anyway, my first idea for a new game where I could avoid the usual conversation involved having a mute player!.. probably the flunky to some other character.

Then I had a couple of less extreme ideas. They are far from whole solutions, and Larroquod may still not like 'em as they use a tree menu, but there’s a couple of different approaches. Note they’re best suited for characters who are either clear on their goals or sociopathic :wink:

  1. When conversation begins, you specify where you want it to head. Whether or not you get the desired outcome, or something like it, or nothing like it, depends on the gamestate, who you’ve spoken to, what you 're carrying if that’s relevant - have you created the circumstances to achieve that outcome with that person? The game checks and then produces the full flow of conversation.

So you only have to do one thing when conversation starts, which is pick a broad direction from a list determined by context, but there are still different conversations possible.

  1. My second idea was like CYOA but divorces you from knowing exactly what you will say.

A problem with a lot of CYOA is it’s hard to be neutral about player options. Some options stick out like crazy.

“Do you wanna say ‘Yes’, ‘No’, or ‘Wait, tell me about the prince of Zanzibar!’”

So I though about having a set of broad approaches to conversation, rather than specific content.

An NPC starts talking to you. You could then choose to respond ‘Very positively’, ‘Positively’, ‘Neutrally’, ‘Negatively’, ‘Angrily’, ‘Flirtily’ etc. You pick your approach then conversation takes off by gamestate and circumstance. Any conversation is prefaced by the same ‘approaches’ menu. What’s neat is you don’t have to specify any topics. They should come naturally from gamestate. Also you don’t know precisely what your character will say before you say it.

This one seems especially good for sociopaths, and I was thinking of making a game where you play a sociopath, but I probably won’t.

This is an excellent observation that I agree with wholeheartedly. Arguments to the effect of ‘But my character should be able to decide that because logicially it should be in his control’ are rarely persuasive to me, because I have never seen a game where you can control everything the character ‘should’ control. The designer of a game chooses to make some subset (and it is always a subset) of possible choices available to the player. This decision on the part of a designer is an artistic decision — IMO it is not really susceptible to formal logic of what the player ‘should’ control. It sounds like we agree on that point but simply disagree on other factors that constrain those decisions on the part of the designer – like what kind of story you want to tell, which is fair enough as far as it goes.

I do think though that you are vastly underestimating, above, what can be achieved storywise without directly suggesting topics of conversation. You are restricting it to very narrow categories but I perceive way more possibilities than that. Possibilities that I can only really ‘prove’ by demonstrating them, so that’s where I’ll put my efforts. Mistake I used to make: letting my passion for debating the issues wander into the territory of explaining all my ideas and letting the juice run out of them that way, rather than saving that motivational juice for implementation. So I’m going to try to avoid that this time and leave myself in a state here in this thread of ‘not having proved it yet’ – which is a good state to be in, artistically. 8)

That politician game sounds perfectly awful to me, even without the indirect manipulation, lol. BTW portraying what I’m advocating as simply manipulating the contents of the room to achieve character effects, is mostly technically accurate (if you discount timers and daemons, etc.) but quite reductive. To the extent that ‘only by physically manipulating the contents’ is pejorative, it would have to turn on conflating the abstract with the specific. But the ‘object’ and the ‘room’ we’re talking about here are just abstract representation of the idea of containment, representations that you can actually use for anything that involves containment, not just literal objects in literal rooms. Which is another avenue I’m exploring in my first (small) game. So, I’m glad you qualified with ‘rarely’… because in my conception there is nothing really ‘only’ about manipulating objects in rooms: that IS the IF language and what I believe is the best scaffold for accessing almost all effects, including most dialogue.

Everything is an abstract representation already, so adding a ‘topic’ level abstraction is abstracting away from the room abstraction. I would prefer to bend and twist the room abstraction than to go there.

Think about what DW Griffiths achieved by introducing the world to the time-eliding cut (in film). That technique made no sense to many at the time, and an argument could well have been made that cuts are OK for a certain narrow set of action stories, but for elaborate character mysteries, one needs to avoid the cut and let the conversation play out more like a stage play. They perceived back then that a certain tool was REQUIRED for a certain type of narrative job, because it makes a logical sense of its own and seems persuasive, but it turned out that logic wasn’t necessary at all because it mistook what is the key factor to enjoying and getting story value out of subtle character interaction, and that factor just isn’t the accurate passage of time. What is the enjoyment factor? That’s the only key question, I believe, not what seems formally logical according to the conventions of the medium.

It’s not a perfect analogy, to be sure (one can’t involve player agency in it for example because there is none in film) but I think the state we are in with IF is still nowhere near what it could be mostly due to leftover assumptions from the birth of the form, and I believe the ask/tell conventions are one of those leftovers. This has all happened before, in film and other media. (Remember how all the early novels were written like discovered diaries? Because the authors were worried that audience needed an explanation for why are they reading this. Nobody worries about that anymore. I can envision a day when nobody will worry anymore about whether they ‘should’ be able to control individual conversation topics in IF.) So that’s just my POV, I’m characterising it how I see it here, I realise that this analogy is flawed and doesn’t prove anything and is not really a killer debating point.

But there are always great logical arguments for keeping things that don’t really work. That’s why in all the early movies we are watching people get into cars, drive cars, and then get out of cars – people thought you couldn’t tell a story that involved driving around without showing it. Audiences would get irritable if they weren’t shown things that are really, pointless to be seen, anyway. Convention is just that powerful. That’s how I feel about making conversation topic choices in a game, these are almost always pointless choices and thus shouldn’t be available to me. YMMV as of course it obviously does.


I should take this opportunity to note that several people debating this here with me have spent a lot of time doing exactly that sort of thing, and in no way do I intend to denigrate the value and artistry of what they’ve achieved. For me this is a debate about the future of a certain form of interaction — not a passing of judgement on anyone and everyone who has put time into that form in the past. The flaws I perceive are in the form, and not mostly in the artists. So their efforts are NOT diminished by my view, and of course what they’ve created will always be an important part of IF history, full-stop.

I prefer a help menu, personally. I am not much of a stickler for not breaking immersion. I see the hint menu as of a piece with the window and the scrollbars. Is it breaking immersion for the player to say, maximise the window, mid-game? Should we, therefore, prevent it? There is nothing pure here which is made impure by a hint menu; that level of immersion just doesn’t exist to be broken, IMO. But I take your point – choosing a topic can be a useful in-continuity way to get a hint – and I agree.

Not that I know of. And though this thread will last forever if it becomes ‘Lots of people suggest scenarios that require ASK/TELL and then Paul says whether he thinks they really do or not’, I’d point out in your scenario that asking people directly about other people is usually the worst way to find out if you can trust them or not. In real life, they tell you whatever you want to hear, and then their actions speak louder than their words. So, if I were to design your situation such that untrustworthiness is revealed through a failed attempt at cooperation with an NPC on an actual physical task, rather than improbably through a clumsy form of interrogation, it would actually be more true to life, wouldn’t it? And it would fulfill, ‘Show, don’t tell.’

But anyway I am not trying to argue that every situation is reducible to action in this way. There are some situations which aren’t, I admit this, so I don’t really need counterexamples because I know they do exist — I just don’t think the above is one of them. And, I would argue, there are a fair bit fewer of them than people tend to assume…

Note that ‘triggered automatically’ is not the alternative I’m proposing. I’m proposing ‘triggered indirectly by relevant player choices which are not suggested topic headings’. Triggering automatically (which I read as ‘according to some timer or predetermined schedule’) is just one of a myriad of ways of fitting that bill. Not usually the best one, either.

As for choosing not to have the conversation, yeah, you’re right about that. That’s why I said I am bit ‘softer’ on things like ‘talk TO’ or ‘greet’ — so that it is easier to choose not to have the conversation at all. That I can see a lot of use for. But then again, you could just not examine or pick up the objects that the other character finds interesting, so then he/she wouldn’t see you as a kindred spirit or whatever and wouldn’t follow you out of the room, and thus you wouldn’t be having continued conversations with them — and that for me would work just as well, as one ‘shared-object’ technique among many.

What I want is to focus the player not on, ‘what’s the magic password for me to say to this character’ and to focus the player more on, ‘Who is this person and what do they value in this environment that we share? How can we connect through this environment.’ Because in IF the environment IS the story, so mediating everything through that is to me a big strength and not a weakness needing redress through a specific conversational object-bypass system, implementing, essentially, a whole other level of containment (people ‘contain’ topics the way rooms contains objects), when there is a perfectly good containment system already there if you just ply it liberally enough. But the more you ply it quite liberally the better it works. I don’t want to give examples — I want people to play my examples, say 4-6 months from now and such. The training game I am writing encapsulates a lot of these ideas: I only intended to do it to learn Inform 7 (big task, requires small test game), but my ideas could not be suppressed — they’re all popping up in there, one by one. 87

However, the game is not ballooning in size — on that I’m maintaining discipline, so far. There will be four puzzles, two object-based, two more about manipulated what other characters decide. So hopefully there will be a nice range in there. There is no specific genre, it’s mostly slice-of-life-with-a-strong-narrative-throughline in the way it feels to play (well, there is a little sci-fi surprise.) Anyway, I can’t prove it all in one short game, but I couldn’t keep my crazy ideas out of the game, either, and remain motivated — so this is what has me thinking about all this stuff, all over again, of late.

I’m actually glad I haven’t really convinced anyone. XD

And, ye gods, I am all typed out. I’ll look at the rest of the responses later and try to be more succinct in reply, I promise. 8)


I would have defined “automatically” as “happening without the player giving a command for it to happen.” If the door closes and locks behind you, that’s automatic, no?

(I know I’m sounding rather negative to your concept; there’s no reason it couldn’t work and experimenting with very conventional parts of the game experience is a good thing, but I really only see this going two ways: A game where the dialogue is just flavor text of no importance, and a game full of frustration as conversations trigger when you don’t want them and refuse to trigger when you do want them. But yes, go ahead and give us a demo and then we’ll see.)