Conversation Topics

One of my testers suggested that I include a “possible conversation topic” command in my latest creation. I have done this once before but normally put any such clues in the hints. What is the general view on this? Good idea or not?

I vote yes.

It depends on the game, of course, but in general I would also vote “yes”. Few things are more irritating than an ask/tell interface and no idea what you ought to talk about. (Or even if you have some idea, you’ll often have to try out dozens of topics.)

You wanna avoid this happening:

ask npc about conversation topics

In my less goofy reply - I guess this is the trouble with ask/tell, and potentially the trouble with giving what you might call hints in that area.

If you give players the topics, it can feel like choose your own adventure but with more typing required.

If you don’t give them the topics (and the topics aren’t well clued - this is the important part), it can be like a guess-the-verb problem, but a worse one than usual.

If testers are telling you they want this feature, presumably they’re not lying. But could their wants be solved with better clueing about the topics from you in the game prose? Or are they the kind of folks who would want this feature regardless? Those might be more important questions.

I vote no.
Not that I don’t agree that it will probably be a big help. But if you’re in a situation you don’t ask the person in front of you: “What topics can I ask you about?”
The idea of IF is the player figuring out what to do or ask and if he fails, well, that’s bad luck. That being said I would of course run a conversation where the NPC (or whatever) lead the player towards the topics.

No, if he fails because he’s trying “Ask George about photograph” and the game wants him to “Ask George about snapshot,” that’s bad game design. And, as a designer, you won’t be able to catch all the reasonable synonyms.

I’d think of the TOPICS command as a sort of hint system, but one that comes up so often that it’s worth providing a special situational shortcut for. Topics are more like verbs than nouns, since they don’t correspond to objects, so a conversation with a topic system is like a puzzle where you have to use a different custom verb for every interaction. You can and should make it so that most people will guess the topics, but you also can and should make it so that people who can’t guess the topics can easily figure out what their alternatives are.* And in this case, since it’s very easy to tell what people want a hint about – the topics they can ask the current NPC about – it’s a good idea to allow them to access it with one command rather than making them go through a whole menu.

In fact, I’d say that with maybe a very few exceptions – Aisle? – having a puzzle based around guessing a topic is always a bad idea. The number of topics you might ask a person about is almost infinite, so for the puzzle to be fair you basically have to give the answer away, and then it isn’t a puzzle anymore. (Maybe you could have one NPC tell you that it would be in the PC’s interest to ask another NPC about a certain topic? Not sure. If anyone thinks they’ve got an answer, that’s great.) If your testers are telling you you need a TOPICS command, you do.

This is part of a general feeling I have, that the hint system (or at the least the walkthrough) is an essential part of the game’s functionality. In IF players can get absolutely brick-walled by a puzzle when they can’t guess a verb, an action, or God forbid a topic. It’s part of the author’s responsibility to make sure that they can still get something out of the game, preferably without having to go online for hint requests.

*I’ve been thinking of a game design where you would have a surprising conversation choice that opens up a different plot branch, and where I wouldn’t want to disclose the topic because it involves the player making a moral choice that I’d want them to envision themselves, instead of having the game inform them that this choice is available. But if i did this, I’d warn the player up front that they could sometimes explore topics that weren’t in the topic list, and I might include “(and one or more hidden topics)” in the topic list whenever one was available.

I wouldn’t want people to regard a walkthrough as an essential part of functionality in a game made now. I think including some form of walkthrough is a design choice for the author in relation to each game they make, not a responsibility (maybe unless you sold the game - in which case it’d be so well designed that they won’t need the walkthrough much or at all?). If the default position is that there’s always a walkthrough, that doesn’t encourage strenuous testing and thought about the game.

If a game is crazy hard so most people need the walkthrough every 5 seconds, you’re barely play the game, and that’s saying something about the game design, which could have been dealt with before release… Unless the whole thing was designed to operate in that impossible way, in which case the walkthrough is now part of the design.

A badly conceived walkthrough (especially just a list of commands) can trash a game experience quicker than it fixes your stumbling block.

It’s probably more realist to want everyone to ship a walkthrough, but I’d rather authors decide what they want to do and then stand or fall by their choices.

  • What I’d enjoy is a box on the game cover with a tick, saying ‘a group of humans was able to complete this game without any hints at all from the author’. It doesn’t matter how big the group was, so long as they did it only with bouncing amongst other people whenever they got stuck. That to me would be a big endorsement of a well designed game… So long as it didn’t take them years or something.

In my personal biases, I have a feeling that if your playtesters (collectively) can’t clear your game without any help from you, you’ve failed to demonstrate something about your game. Except in a purposely crazy difficult game. But maybe even then, too.

You’ve got to find some way to clue players in if you’re using a topic-based system of conversation. Especially if you don’t have an insanely large system in place or the conversation is non-optional.

I’m showing my own social anxiety issues, but I often do think about what topics I can discuss at any given time, even in casual social settings. And if game-me is conducting an investigation, say, I’ll certainly have a list of topics at hand. I think of TOPICS as “think, in-character, about what I might want to say next”, which pretty much everyone does in conversation, especially conversations where only about 1% of your carefully selected options are going to yield any sort of progress.

If you’re not going to have TOPICS, you need to have something else. Your play testers are struggling: what are they struggling with? Is it a lack of implemented topics, or are they unclear what the goal is with the conversation they’re having? What are they missing?

Most importantly: how are you going to help players who don’t hit an essential topic or are missing the topics you’ve implemented?

Re: walkthroughs - I prefer a robust hint system or at least a written out walkthrough (sentences rather than a list of commands). I think it’s a necessary part of functionality; one you hope won’t be needed much, but one that will dramatically improve the play experience for a significant cohort of your players. That’s based on a vague goal of having lots of different people play and complete the game as far as they want to; other writers have other goals, of course.

Thanks for the advice people. The trouble is I’m not really sure. My game has a hint system, which eventually will give away the answer basically. The particular point in the game was odd as well. An office, with a desk. On the desk a framed photograph of a woman. One NPC in the room. So I would ask the NPC about the desk, photograph, woman, wife, girlfriend, mother etc. Could also break the photograph to see if it generated a response. But, that is a different line of discussion!

If I did include a TOPIC command it would be experience based - only listing the topics you should know about at that point in the game.

Well, in a way this illustrates my complaint – the number of things you can ask about is so open-ended that it might be hard to hit on (one of the) correct topic(s).

Maybe in this case you could have the NPC start talking when he notices the PC examining the picture?

I had a similar dilemma in my game. I implemented a very basic ASK / TELL ABOUT dialogue system. And from my testers I was able to see lots of different things they tried. So I fleshed out a lot of synonyms and various other phrases to be handled either with synonyms or occasionally with an UNDERSTAND AS A MISTAKE command. Nonetheless, some testers still felt they had to guess the topic. And as much as I might think things will flow in a certain fashion with the player picking up on phrases leading to deeper exploration of new topics, it doesn’t always unfold that way for the player. So I did end up implementing a TOPICS command in addition to the full online hints. (I also implemented a more limited HINT that just tells you something about the current location.) While a perfect game might not require it, I felt that it’s up to the player to decide if he or she wants to use the TOPICS command and I’d rather they not feel totally stuck.

I don’t know how other people implemented TOPICS but mine is quite simple. I look to see what NPC is in the room and if there’s just one, I list some topics. (If there are none or more than one, I am a bit more generic in the topics suggested.) If there are some items that have been discovered, I add a few of those. (And some topics like “whereabouts” or “alibi” are also applicable to other NPCs.) So while it’s not a completely static list of topics, it’s also not a full blown dynamic system. Basically it was less than an hour’s worth of coding for the 5 NPCs and other cases.

I wrote a bit about the use of TOPICS and keyword interface here: … rface.html

I just played Maher’s King of Shreds and Patches, and made full use of the topics command. I hate the idea of forgetting to ask about one key topic that would have opened things up (and having to traverse the map again to get it, assuming that was even possible). Too, there were enough characters and plot threads that remembering who was doing what and knew about what was a non-trivial task.

Really, the biggest thing for me was that I wanted to know The Next Thing. The story sucked me in, and I wanted it to progress without delay and not TOO much fumbling.

In less complicated games, I think you could get away without an explicit topics command, but the player needs to be cued very well on what lines of inquiry might be productive.

I just had an idea… what about bringing up a few topic suggestions when the player asks about a non-implemented topic and gets the default reply? The advantage to this might be that it could, if the author wanted, be done without non-mimetic (I think that’s the right term, outside of the storyworld) messages. In a few situations, perhaps the NPC could reasonably tell the PC what topics are open for conversation; more likely, the PC would refuse to talk about the non-implemented topic and tell the player some things that he or she thinks are more useful topics. Of course, a “TOPICS” command could also be in-character, but this might feel less like “cheating” or using a hint system.

So, it would look something like this:

You don’t want to waste the Captain’s time in idle chat. You have a specific mission to accomplish, after all. You realize that captain might know something about the tidal anomaly or the pirate raid if you ask him, and you think it might be worthwhile to tell him about your discovery of the secret map.

I like this.

I love it.

I liked the topic menu in Pytho’s Mask, but in general I prefer not to have topics suggested outright. I found the menu system in Photopia to be a bit annoying.

Two games that made good use of ask/tell without topic suggestions were Lydia’s Heart and Worlds Apart. In both, the conversations were either optional (providing hints but not directly advancing the plot) or relatively easy (there was no complicated branching path of conversation that needed to be traversed).

Conversational puzzles tend to bug me in IF. In real life, I rarely achieve any difficult goals by choosing exactly the right thing to say at the right time to the right person. When a real conversation happens, everybody usually gets to say everything they need to, but the information can be ordered in lots of different ways. For that reason, I’m against the kind of complex AI-based state machines that are used in some conversation-heavy multiple-endings IF works (although I do have great respect for Emily Short…)

I’ve seen some works where there are 2 or three suggested topics of conversation at any time, which seems like reasonable range of choices. In my WIP, I have exactly one “suggested topic” at a time, but it’s presented something like this:

We’ll see how it works when I try it on actual testers.

i like the use of the think action to give a bit of a nudge. i was doing that in a WIP ive since back-burner’d.

I’m not planning to document “think” explicitly, since the game will always tell you when a new thought pops into your head. But I thought it might be a friendly use for an otherwise problematic verb.

The obvious risk of what I’m doing is that conversations might seem completely railroaded. But I’m attempting to keep them short, and not put any significant puzzles in them. Although the player will always be able to bring up a different subject, the main choice presented will be whether to speak or not.