Gamebook store

Thursday 28 April 2016

Gamebooks: the value of doing it with dialogue

Eighteen months back, Leo Hartas talked me into starting work on an interactive story app. Leo is a very persuasive fellow, and it sounded such a beautiful plan the way he told it. I’d write the thing, Leo would do the artwork, and his coder friend would put it together in his spare time.

I really should have known better. Spare time is pretty much a mythical concept anyway, and so the chance of such a project ever happening decreases exponentially with the size of the team. After six weeks and two Skype calls, the coder admitted he was too busy and Leo got a contract to illustrate six kids’ books. The project went quietly back to bed and set the alarm clock for never.

Still, all experience is useful. Some of the writing made its way into a novel called The Mage of Dust and Bone that Jamie and I may yet finish. And I had enough fresh insights about gamebook app design to fill this blog post. So not a total loss.

The app, which was going to be called Winter’s Rage, was a sandbox adventure based on the Christmas scenario here a while back. The map would serve as the top level of navigation through the story, as in Fabled Lands. Nothing new about that; it’s been a staple of CRPGs from Might and Magic right through to Sorcery.


‘When you have an encounter, you’ll drop down from the map to a location screen that will be mainly text.’

‘I could do illustrations for each location,’ said Leo.

‘No. Doesn’t work. See, we’re used to using a map in real life without thinking about it. Our brains have the subroutine that means we look at where we are on a map and that’s part of our seamless word-view. But as soon as you put a picture of a church, say, in front of somebody who up till then has been picturing the world as a map or via text, suddenly they’re thinking, hmm, that’s not a real church, that’s a drawing. So then you’ve broken suspension of disbelief.’

‘But I thought you said you'll be wanting little mugshots of the characters’ faces for when they’re saying something?’

‘That’s different. When the brain is used to interpreting images symbolically – a figure on a map, a face next to some dialogue – then artwork doesn’t pull you out of the story. Given that we have to have text, simply because it’s cheap, we need to let the text be the player’s main “world rendering medium” and any artwork has to conform to that design principle. That is unless you can find a million dollars down the back of the sofa, in which case we’ll do it all in a 3D environment with audio.’

‘Text it is.’

We came up with a screen template for locations like this:

OK, OK, gimme a break - I'm the designer, all right? Leo would've done the art. Anyway, at the top you’ve got the location name (The Bank Road in this example) and under that a brief description that sets the scene. Then you’ve got the faces of the characters who are here. In this case there’s a Blind Man who you just met at the start of the adventure and who is accompanying you along the road. Your answers to his questions are creating your character at this point – eg he begins by asking if you’re familiar with these parts, and you can answer either ‘I’m an outsider’ or ‘I grew up not far from here’.

But I digress. The point here is that you’ve just had an encounter, that’s why the view has dropped down from the map level. The encounter is with the Robber. So then we look at the pane below, which is the dialogue. This is where the action of the story gets presented. Why? Because the key to keeping the player’s attention is writing in the moment. That’s not new either; it was invented, or at least popularized as a novelistic technique, by Samuel Richardson in the mid-18th century. If you’ve read my Frankenstein app, you’ll see the same technique in action throughout. The entire text there is what Victor Frankenstein is saying to you, so his words must carry all the narrative, rather like in a radio play.

The advantage of placing the narrative emphasis on dialogue is that readers of an app will skip descriptive text. Description is less compelling to an untutored eye even in a regular novel, and when you’re leaning forward waiting for the next decision point, the temptation to scan for surface meaning may become irresistible.

Not so in the case of dialogue, because we’re attuned to care about other people and what they say. Arguably the main reason for these big, energy-hungry brains is to interpret the nuances of meaning in speech.

The way we planned to do it in the app, the dialogue would appear one speech bubble at a time, with a beat between them. You could read and re-read it at your own pace, obviously. The beats were there just to reinforce the sense of them speaking in real time, rather than everybody’s dialogue appearing at once like on the page of a book.

So the Robber says, ‘Hand it over. All your money.’ And it turns out your blind companion also has something to say: ‘Gar John’s-son, I know your voice. Have you turned to thievery now?’

(Later in the adventure, you will typically be travelling with a companion – more about them in a minute. For example, if your companion was Fosse the hunter, in this encounter he’d now chip in with: ‘Huh. Since when wasn’t he a thief? Five years old and I caught him taking rabbits from my traps.’)


Then at the bottom of the screen, under the dialogue for the encounter, you’ve got the DECIDE tab. When you tap on that, you’re presented with your options. Why not display them right away? Extra unnecessary taps/clicks are usually a bad idea on an interface, aren’t they? Yes, but here it’s to stop your eye just scanning straight down to the bottom of the screen for the options. It keeps you in the story.

If you tap DECIDE, in this case you’ll get two options to choose between:
  • ‘It’s not gold you’ll be getting from me, it’s cold steel.’
  • ‘I don’t need to fight you.’
As often as possible, like there, an option will be a line of dialogue. Say you choose not to fight. That dialogue gets added to the scroll underneath what’s been said already. Then, after a beat, the Robber will reply to you:
‘I don’t need to fight you.’
   {set #Spared_Robber = true}
   // #Spared_Robber is an inline conditional not visible to user
   // sets reminder that Gar Johnson may be encountered later
Robber: ‘I’m starving. I was sick. Couldn’t get no work.’
And your options now:
  • ‘Take this coin and buy yourself some bread.’ { if #Status_Noble == true }
  • ‘Winter’s hard on everyone. You’ll survive.’
  • ‘Come and see me at the manor house. Maybe I’ll find a job for you.’
Options can be conditional, as in the example above where the option to give the robber a coin is only available if the player is an aristocrat.


You pick up companions at the manor house, which is the player's base of operations for the game. You can only pick one companion to accompany you at any given time. You can change companion when you return to the manor house, though sometimes they may be absent on their own errands, depending on the adventure timeline.

Companions become more loyal to you over time, assuming they see you solving problems and showing good leadership. A companion who is more loyal to you will volunteer more personal information (possibly unlocking backstories) and also uses their skills more effectively. Therefore it makes sense not to switch companions too often. Balance that against the need to have the right companion with the right skills for specific tasks.

When you return to the manor house you can show any clues to the steward or the NPC companions you didn't pick up - so you can get the benefit of advice from any of those four listed with a delay, but only the one companion you pick to go with you will notice things, prompt you during investigation, help in fights, unlock subplots, etc.

A possible mechanism for giving the player hints is that when you’re at the manor house, companions will talk to each other (‘You won’t believe what I saw up on the heath, Sir Werian…’) based on their preset relationships, and if you listen in you will get the benefit of their theories.

NPC companions add an element of communitas and emotional grip to an adventure game - a discovery I made by happy accident when writing Down Among the Dead Men, though I guess the germ of it was there in the interactions with the faltyn in the Blood Sword series. As a rule, people are way more interested in character and the development of relationships than they are in facts and the development of plot. Given that Winter's Rage was to be a gamebook app constructed almost entirely through dialogue, and an investigative adventure at that, having a foil for the player to interact with was essential. Not only did the choice of companion on each mission mean a set of skills and insights that would customize the experience, the companion also gave me as writer another pair of eyes and another voice to interlocute the world for the player. And, as this is an adventure with a ticking clock, interjections from the companions can be used to ratchet up the tension.


Text is inexpensive, but there are some tasks it doesn't handle well. Artwork is useful for the top-level navigation of the adventure (ie the map) and to depict the faces of nonplayer characters. But don't be tempted to use more than that. Presented with a little judicious artwork, the brain interprets it symbolically; too much and your "gamebook app" becomes a broken CRPG.

Also, use dialogue as much as possible in place of descriptive text. Even in the example above, in the final version I'd probably have lost that descriptive line "a robber steps out" in favour of a companion saying, 'It's a robber!' or 'Now who's this?' or 'This guy looks a bit shifty' depending on which companion it was, the time of day, that kind of thing. After all, you know you've dropped down from the map level because an encounter was triggered, and you can see the guy's face has appeared in the mugshot pane. You don't really need a stage direction to tell you what's going on here.

In order to get maximum mileage out of dialogue, focus on writing to the moment. Listen to radio plays. Read how it's done in something like Pamela, Riddley Walker or the Frankenstein app I mentioned above. Anything you can let the dialogue carry, do so. Strip down descriptive text to the barest scene-setting. Don't tell when you can show. There's nothing new about any of this; it's just that it hasn't often been applied to interactive literature before.

*  *  *

Writing this, I’m thinking it’s a shame we never got to do the game. It was to be the first in a series of interactive adventures set in the world of Legend. If I’d only had a coder with no wife and kids and re-enactment hobby to eat into his leisure time, eh? Ah well, it’s water under the (half built) bridge.


  1. You know, Dave, I always thought that the biggest difference between your gamebooks and pretty much everyone else's is the dialogue. Most gamebooks relied mostly on description for simplicity. So much so that, when authors did an attempt at dialogue, it was almost invariably cheesy, if not downright cringey.

    But as much as dialogue makes everything more colourful and interesting, I think there's a catch. How much of the main character's dialogue should be explicitely written by the author? After all, the reader IS the main character. Putting too many words in their mouth could actually draw them out of the narrative. There's some dialogue in Heart of Ice that did make me feel incredibly cool... only to make me think afterwards, "I would never have had the wits to say that". Suddenly the character was indeed cool and witty... but they weren't "me" anymore. It's interesting that until then I always considered dialogue in gamebooks, as cheesy as it could get, as perfectly expected and in character, but when it tried to be more than that, it could disconnect me from my character. I don't know if this says more about my poor imagination, or about the pervasiveness of dialogue tropes in popular entertainment.

    1. I've felt the same playing something like The Witcher, Efrem, so I agree with you there. But what I'm mainly talking about here is carrying the story using NPC dialogue. That's the reason why I included a companion for the player; the companion will have the more highly characterized dialogue while the player-character's own comments are deliberately kept fairly bland.

    2. "After all, the reader IS the main character. Putting too many words in their mouth could actually draw them out of the narrative."

      I think that this could be solved in the character creation page. Suppose to be able to choose 5 adjectives to describe your character personality.

      Like: conceited, risk taker, trickster, misogynous, defiant.

      Now you have a nice start point to build dialogues that fit your character idea.

    3. Even if you do that, you can see that writing the player's dialogue for them has a distancing effect. Say they pick misogynist and you give them the line, "Shut it, bitch" (not in a medieval adventure, admittedly) but the player didn't intend the character to be crude & rude like that. He just meant he'd be condescending towards women and say things like, "With respect, ma'am, I don't think you know the first thing about men's business." The point is, you're putting words into their mouth, and it might be a necessary evil but it's still never going to be quite what the player themself would have chosen to say. And that's why it's a problem.

  2. As an extension to Efrem's comments, I usually get frustrated with CRPGs that present dialog options for my character, as they rarely present an option appropriate to the direction in which I would like to steer the conversation. However, and whilst I'd never really thought about it before, I do find CRPG dialogue far more engaging than other elements of CRPGs - I guess for the reason that it's usually the point in the game at which a decision I make might actually _matter_. Wandering around Skyrim can be all very well and good, but it's only when your character talks to someone that quests are revealed or the story progressed.

    And I sympathise with your coder-friend - I, too, have far more ambition than I have either the skill or the free time to realise. I've been meaning to write something in a similar style to the 'Life of a Wizard' for over a year, but have still yet to create more than the briefest outline sketch of how it would go. When it does finally grace the Internet with its presence, your article has made me think about making it far more dialogue-orientated.

    Thought-provoking stuff, as always.

    1. It's particularly a problem in CRPGs because you expect to hear your character say *something* - yet the options given are rarely a good fit with the player's own preferences. The player can explain that by thinking, "OK, that's what Gerault would say," but that merely invites a disconnect between player and character.

      We have a little more leeway in a gamebook app, though. Because this is at least partially interactive literature, you can use reported speech for the player-character - or just present the unadorned actions with no dialogue. Because the player is internalizing their character's thoughts, a lack of dialogue there isn't so jarring. As long as the NPCs are all chattering away it works fine.

  3. 'And what is the use of a book without pictures and conversations.' Alice in Wonderland.
    Would love to see the idea in this post implemented somehow - perhaps just mainly with the conversation choice aspect to minimise technical requirements?

    1. To be honest, Adam, the project is about 20% coding, 15% art, 5% audio and 60% design/writing. So it wouldn't save a lot of work to make it text-only - and it would lose too much, I think.

  4. Awesome article, which hits upon a lot of the beats that I would like to try (or see tried) in IF. I'm not entirely in agreement on the role of illustrations - I feel that they serve their purpose well, especially when used as in visual novels (backdrop art) or to set a tone/mood - but you have some good points.

    And you certainly have a point wrt dialogue vs descriptive text. I've enjoyed reading a lot of the new "renaissance" of IF on mobile devices, but the number of times I've just flicked through paragraph after paragraph of descriptions to get to the choices have been innumerable. I can barely restrain myself on first reading, and I never do if I replay. Stories that actively force me to "page" through texts just annoy me.

    What you're describing is pretty much the visual novel approach - dialogue-based story beats, intensely focused around characters - except that VN often tone down the amount of interaction. Study modern CRPGs, and it's pretty clear that it is the character's that create the classics. Mass Effect won't be remembered for its gameplay excellence or stunning visuals (although both were great for their time), but for Garrus, Talih'Zora, Liara, and Wrex. I don't remember much of Baldur's Gate anymore, but I'll probably carry Minsc and Boo with me till I die. As you call it the emotional grip of the adventure is in direct relation to the communitas that the reader feels to the story's characters. IMO, a lot of the success of inkle's 80 days is founded in the same aspect (the player's caretaker role for Mr Fogg).

    In short, I think you're right in seeing this as the direction that IF will eventually go - with or without illustrations.

    For my own part, I'm fascinated by the idea of narrative combined with procedurally generated characters; I blogged about this some years back ( I feel the need to blog some more on the topic come upon me, but mainly I think that people want to care about characters, and thus it should be possible to create random characters that the player feels a close connection to, even if they are not written specifically for the story. Possibly at some cost in communitas (though I'd argue that this cost would be minimal), but with promise of creating a much more replayable and potentially interesting narrative.

    That's the idea I'm working with for the narrative engine in "Pirates and Traders 2" (recently re)built upon inkle's open-sourced ink scripting language. There's a long distance to go (and not least - a lot of writing) before that gets to the point where I'll know whether that works or not, but the idea is to hit those notes with lots of dialogue, terse descriptions, and full focus on the characters.

    +1 to the guys who think it's a shame that the game didn't get made, although I'm really looking forward to seeing how Frankenstein Wars turns out. I'd say though, if your thoughts turn back to the project someday, consider dropping me a line? It's not something I could take on right now, obviously - but perhaps 12-18 months down the line, the P&T2 game engine may be sufficiently settled that I could build an app for something like this without too much effort. At least that's the idea.

    1. I'm interested in procedural characters too. Back when I was designing Dreams (a soap-opera type Sims game set in a small town) for Elixir Studios, the characters' behaviour was decided by mood and relationships, and I had a basic psychology system based on how easy it was for a character to transition between states and where their baseline was - the state they drifted back to over time. Having done that, I could make a set of customizable characters who were the "guest star" NPCs, but also generate as many unique "extras" as we needed. Winter's Rage would be similar - each of those named characters being based on a skill-set, attitudes and a knowledge base, which implies that other characters could be created out of the same attributes. I wouldn't write the encounters as special cases, instead it would look for the relevant attribute and that would decide what the NPC says or does.

      The thing that was already starting to make Winter's Rage considerably more complex than something like Fabled Lands or 80 Days is that it didn't consist of micro-quests with a smattering of cross-connection. It was a detailed whodunit storyline in which every event would have a knock-on effect. But then, that's why I was interested in doing it!

      Good luck with P&T2 - that sounds an interesting project too. I'll watch its progress with interest. And maybe in a year or so we'll work together on getting Winter's Rage back up and running.

    2. I think it's very interesting with psychology systems for IF/role-playing games; though the feedback mechanisms (letting the player understand why a character reacts the way they do) is a challenge.

      P&T2 characters combine a World of Darkness like skill system with items and "traits". The latter basically covers all types of quality attributes; both personality traits, titles, opinions, etc. Still working out how to really exploit all of the elements for good narrative, but one part of it was definitely to use the traits to determine behavior and dialogue, like you describe.

      Winter's Rage sounds like it would be a very interesting project. So much to do, so little time...

    3. The skill system I was thinking of could form a whole other post. I had a couple of design goals. First that a failed skill didn't simply block the story. You might fail to spot a clue, but that could trigger an interaction with an NPC (say) that would move things along in other areas - trust, hints, whatever. And also that I didn't want the player to see skills as numbers. I think just fine in numbers myself, but I never feel like House of Cards would be improved by seeing Frank's SmoothTalk or Connive skill level flash up on screen. In a face to face RPG, skill numbers are dealt with by the players largely subliminally - they can be both in the reality of the story and spectators watching the dice rolls. But for an app like this, I want nothing that moves the player into a "gamer role" rather than a "character role".

    4. That does sound like a blog post waiting to be written (nudge, nudge, wink, wink).

      I completely agree that skill checks in these kind of interactive stories have to be non-blocking; a failure that blocks the story or cuts off content entirely is just frustrating. Much better if things can then take a different turn. I also like to provide the player with multiple options based on different skill groups, so maybe if the player cannot smoothtalk their way past an obstacle, they can sneak or strongarm their way instead.

      I find the numbers issue a tricky one. It's certainly better for immersion not to have to be reminded of the mechanics of the game, but on the other hand, one needs to communicate clearly to the player the risks, consequences, and probabilities involved in what they are doing. Lacking the feedback loop to take intelligent decisions is also immersion breaking - and having numbers to work with can help with that. I like how it works in King of Dragon Pass, with the caveat that things can at times get very opaque, which not only makes the game frustrating to play, but also very, very hard for new players to get into. Which could be unfortunate.

      So I'm not sure there is a one-size fits all solution for that; I feel it really depends on the game and the mechanics.

      I really think the tendency in some gamebook apps to show the dice rolling is misplaced, though. TMG have had success with this approach, but I always end up feeling that if I wanted to play a replica of a gamebook, I'd rather just pick it from the shelves. Though maybe I'm just old... :D

    5. Looking back at my notes, I see that numbers did play a part. When the player was presented with the option to use a skill, the system looked at their skill level (which to them appeared only as a qualitative description) and at the task difficulty, and they would be told how many story points they'd need to use to be sure of success. They'd have the choice to spend fewer points (because they only recover slowly), but then all the points spent might be wasted. That part was very definitely still a work in progress, though, so it might well change. One thing I'm quite sure of: I'd never use virtual dice. I absolutely loathe them. They seem to me to replicate the most pointless part of the experience, like having paper-texture pages that rustle as they turn. If the system has to turn on just a random result, I prefer Fallen London's success bars. Dice just distract from the story. Well, that's me :-)

    6. Ah, a kind of fate points mechanic? Sounds interesting. And I totally agree about the pointlessness of replicating the physical aspects of the experience.

      I'll reiterate my offer: if you get around to a point where you would like to look at possibilities of doing something with that project, do consider contacting me. I have limitations on what I can do with MicaBytes - one person, spare time, lots of ideas (always) - but it rarely hurts to talk.

    7. And yet I am disdainful of fate points in face-to-face roleplaying, because it yanks the player out of character into an authorial rle :-) Still, I guess I can rationalize the Winter's Rage system as representing concentration and fatigue. Er... hand-waving... er...

      And for sure, Michael, let's talk further. This could be a project for Kickstarter, which would generate some funds so we could focus the time it needs.

  5. On a different subject, what are the chances of getting Way of the Tiger 8, Descriptive Personal Noun!, anytime soon. Soon being within the next two or three years?

    1. I thought that series had ended now, but if any more titles are going to come out then you'd need to ask Megara about it.

    2. PS: I checked with Jamie and there are no plans for more WOTT books, at least as far he's aware.

  6. I remembered, or thought I remembered that there was some discussion of the possibility for more WotT books. Redeemer words as a series finale but there plenty of other stories that could yet be told about Avenger.

  7. Ron Newcomb had a recent post on Gamasutra that adds to this discussion:

    He is mostly talking about "pure IF", in the sense of procedurally generated stories using AI-driven NPCs. The Full Chris Crawford, as we used to say. That was what I was working on with the game Dreams at Elixir Studios, which is mentioned in passing here:

    But a lot of Ron's points are relevant to gamebook writing too. And if all this is of interest, you might also want to revisit the earlier posts on the role of prose: