Conversation Trees: Necessity or Option?


One of the decisions for which OtherSide has been taking some heat is the choice not to implement conversation trees as in the original Ultima Underworld games.

Let’s poke at that a little more deeply. Thanks to Flug for starting this out (in a different thread):

May I beg to differ with your begging to differ? :wink:

I don’t think dialogue is crucial. I think aesthetics are crucial, of which story is one important form, for which dialogue is one useful tool among several.

A dialogue system for “talking” with NPCs is just one means to an end, not an end in and of itself. It would be a mistake for game developers to think a conventional dialogue system is their only option when trying to convey information about the world through people. Among other things, we wouldn’t have System Shock if that were the case, but the objection runs much larger than just one game that capably communicates story even without the usual conversation trees.

Certainly the conversation tree model is a useful and effective tool! NPCs can directly tell the player character interesting or useful things about the world; the player’s conversation choices can help define that character for roleplaying; there can even be mechanical effects and gameplay consequences based on the player’s dialogue choices. Developers generally don’t implement a conversation tree system out of blindly following convention; they do it because they think it will deliver gameplay value for the kind of game they want to make.

BUT that doesn’t imply it’s the only way to get value. And value has a cost. For a given game, there may be some other way to deliver story information and support character-building choices (and consequences) that works better for that particular game, or that has a lower cost to stay within resource constraints. Haven’t we all talked about products whose features considered cool today were invented out of sheer necessity? At some point, every game feature considered a convention today was a risky experiment someone took a chance on.

Not every experiment to deliver story and character-building in some other way will work. Some will turn out to underperform for their development cost. Making games isn’t pure engineering; there is art to it. Someone who can always guess correctly what art will be universally acclaimed by the public should feel free to second-guess other creators; otherwise I think we are wiser to give game devs some breathing room for testing alternatives.

TL;DR: There has to be some leeway for trying different solutions for all kinds of game design needs, including sometimes not implementing conversation trees. Even if some concepts wind up not working out, it’s important to support a culture of thoughtful innovation in game dev or we gamers will get nothing but cookie-cutter experiences… and then we’ll complain about that.

I don’t know if I’d say the “Saurians speak context-sensitive phrases at the player when approached” experiment was a success. I do think it’s important to support developers when they consider trying something different.


It seems like everyone on this forum has their own pet fundamental design feature from UU that any successor needs to be a “real” Underworld game (aside from the basics of the immersive sim, which we all seem to agree on).

Dialogue trees are mine. Without them, the game ceases to be a simulated social experience. The characters who inhabit the world lose a lot of their personalities and the player loses the one way they had to directly interact with those personalities.

Also, I think the System Shock games only benefited from the lack of dialogue trees because they’re horror games. They make you feel alone. That’s the only reason.

Removing them made the game into a fundamentally different kind of game, in my opinion.

Now that’s not to say I’ll never enjoy Ascendant for what it is if it gets cleaned up and ready to go. I might. But I’ll never see it as Underworld 3.


I heard this was more due to the technological limits of the time, they couldn’t have NPCs because of the engine or some kind of cap on processing power or hardware. Someone at Otherside can confirm?


I don’t remember if it was driven in part by tech or not, but I know there was a design-driven push at Looking Glass away from dialogue trees independent of that. And I get it. It’s not a very elegant piece of design. It’s an extremely limited, clunky workaround of the fact that we don’t have a good way to talk to NPCs.

But it’s a lot better than nothing.

No one pretends that dialog trees are perfect or all-wise. Indeed, it has been said that dialog trees are the worst form of NPC interaction except all those other forms that have been tried from time to time.

There is a kind of game - the “Immersive Simulation”, which is very much like a first person RPG, but removes from the game the essence of feeling of being part of a society. These games can be good and popular and spawn a whole successful series. They seem to fall into three groupings:

  1. Half Life, Max Payne, STALKER, Bioshock, Far Cry, and other FPSs; everyone knows that you don’t get social interaction in a Doom clone. Nobody expects it, most don’t even want it. A linear shooter with some cool weapons and abilities, and maybe a few physics puzzles and interactive objects for “emergent gameplay”.

  2. Assassin’s Creed, Thief, Dishonored, Tomb Raider, GTA; your separation from society is clear just from the game’s name. You go through the game kinda hoping that at some point you’ll see a friendly face you can talk to and relate to, but also kinda know that you probably won’t. You’re playing a linear shooter with exploration and puzzle elements, and if you’re lucky some character-building, inventory and other RPG elements.

  3. System Shock, Arx Fatalis, Fallout 76, and Underworld Ascendant: people expected an interactive world, and were uncomfortable to find that it was just an immersive sim, a world with some interactive elements but no real interaction.

There is another kind of game - the “First person RPG”, where you DO feel part of a society. Again, these can be good: Fallout 2. Ultima 9. Elder Scrolls. Ultima Underworld. Deus Ex.

FPRPGs let you interact with the items in your inventory, using them together; and with items in the world, crafting and repairing things; and with people in the world, choosing how you want your interaction with them to go.

Even though these FPRPGS are where I have spent far and away the vast majority of my gaming life, I still find it really hard to define the precise difference between a FPRPG and an Immersive Sim.

But dialog trees, if not the critical element, are certainly a reliable symptom of when a world falls out of the uncanny-valley of “I’m alone in a dead world” and into the comfortable zone of “I have allies in this alive world.”


I posted my thoughts in another thread, FlatFingers, but I’ll condense them here: Dark Souls, Silent Hill, God of War, Half-life, System Shock, Elder Scrolls–all had different ways of telling a story. None of them relied as heavily on dialogue as on other storytelling contrivances such as narrative symbolism (Dark Souls, Silent Hill series), Cinematics & QTEs (God of War gets a free pass here, it’s such a great game despite its console design), scripted sequences (Half-life), audio logs (System Shock 2), books and lore (Elder Scrolls), ad nauseum. I don’t think the future of dialogue is in stiff menus, or in scripted sequences. The truth is we want something that closely simulates “real” interaction, with all its nuances, such as a body language, vocal tone, etc. Cyberpunk 2077 is on the right track–when you move within a radius of a character, they will start talking to you, and if you move out of radius, the dialogue is stopped. These kind of logical nuanaces will inform how dialogue is delivered in the future–not stiff menu systems (menus are a very poor way of delivering information, as they do not make for an interesting gameplay layer), or linear set pieces. That is my $0.02.


Computer games, and especially RPGs, are interactive. There’s no other way than menus of some kind to choose how to socially interact with people.

Sure, you could not display the conversation menu, and use hotkeys, so hit 1 for a funny answer, 2 for an aggressive one, 3 for the flirty one… this just a global, static, invisible menu that cannot give context dependent options and is less user-friendly.

You can’t really “ditch” conversation trees until someone comes up with a better system for the user to drive a conversation with characters, instead of getting spoken at by talking signposts.

Bethesda, for all their millions, hasn’t managed to do better: they’ve played with circular instead of linear menus, and made plenty of talking-signpost games to show they get the concept: but there’s no real innovation that can really happen in the RPG space until perhaps natural language speech parsing comes of age.


“Just” an immersive sim?

As one more data point here, I wasn’t uncomfortable in the least with System Shock; not only was it highly immersive, I still consider it the greatest game (for its time) I have ever played… and I played my first computer game in 1973.

“Just” an immersive sim, indeed. :o

It’s not much remembered now, but Façade, which featured a text parser for talking more-or-less naturally to characters, was released in 2006.

I appreciate the points you made further upthread, as well as those of the others here. (Still waiting on you, Flug! :wink: ) I’m just noting that there have been games, from Adventure (Crowther & Woods) and Zork (Lebling/Blank/etc/Infocom) to some of today’s Interactive Fiction games (such as those using Inform) that use some form of free text input to interact with the world, including characters. Natural language input is certainly not what one would call a popular choice, and I agree with your closing comment that parsers aren’t perfect. But they’re not awful these days, making this a valid alternative to “pick-a-sentence” conversation trees for certain games.

And that supports the larger argument I’m making that gave devs do not have to get stuck thinking that the conversation tree is the only option, and that they should – must – be free to innovate if doing so is right for the game experience they want to deliver.

What's your "Best game for it's time, maybe ever"?

What about a text parser where the player can type their own replies, ala Wizardry 8? I also liked the “keyword” system of Morrowind, that conjunct with a text parser would do dreams as far as making dialogue options more broad.


Heehee! I’m afraid I’ve put a few noses out of joint by dissing both SS and Arx here, it seems :smiley: I wish I could see in them what others see, because in both cases, all I see is an engine and unrealized potential, with a single linear story and no sidequests or characters or object interaction. They’re fine games, certainly, but I enjoyed them no more than Ultima 9’s dungeons, and for the same reasons.

I know what you mean about having a “greatest game for its time”: for me, Underworld is that game. There has never, before or since, been such a vast leap in gaming. Every RPG since has been influenced by them, and for the most part has fallen short.

But what am I missing, with System Shock? Wait, no, that’s totally off topic for this thread: I’ll go start another one for it ( ) :smiley:

This is exciting :slight_smile: I keep forgetting we’re already living in the future!

I have not played any of the games you list, and now I shall have to keep an eye out for them.


You’re not the only one that found Arx Fatalis only moderately interesting. I managed to break the game, and never finished it.


Again I’ll have to be brief until I have more time…but it’s not only ‘story’ or dialogue trees in isolation…it’s about characterisation, in both senses. Interaction in UU is part and parcel of plot, back-story, pacing, task-setting, and the processing of experiences. Dialogue trees are not just a means to an end, but a reward in themselves, conveying wit, personality (society), hinting at further layers…as wel as all the usual game-webbing-and-narrative-glue. Dialogue works both as mechanism, and story…not just a sterile info vehicle…and let’s not get in a bind about precise method, spoken or written etc (that said, other people’s voices have proved the headache I thought they would. Words are generally cheaper and the better way to go. And doesn’t Mr ‘In-house’ Russell deserves a rest?)


I think on a general note, I agree that there may be some distinction between an interactive world and an interactive environment.

i would agree that for most people, the narrative and your relationships with other characters changes a game from an “environment” setting to a “world”. It’s nice to see that individuals have their own motivations, challenges, and lives out of your interactions. I’m finally picking up Divinity Original Sin 2 again (a masterpiece!) and I’m really enjoying it, even though I tend not to lean too heavily into dialogue-trees. What I appreciate about the game is that it lets me freely make actionable decisions (I can ALWAYS have the option to murder, even a friendly NPC, at the risk of closing off a plot point about them forever, etc) while also being locked into certain dialogue options (curse you, societal rules that dictate I can’t just ignore people when they talk to me!)

I think the problem with narrative implementation for many games is that it either feels too forced or easy to miss. Either I’m bioshock infinite-ing with audio logs during loading screens and I’m forced to listen to someone yell at me, or I’m able to entirely skip past important characters because I got distracted by a dog in Divinity. I think I’m a bigger fan of the latter; it makes exploring the world more rewarding, which is something a couple of you have been getting at.


This is big for me because it speaks directly to the business of making computer games: how do you justify building a game with content that players may never experience?

With a designer hat on, it’s lovely to imagine creating systems whose complex interactions may lead to content that only a few creative players will see… but as a person who’s responsible for spending someone else’s money, how do you tell those financiers that you’re approving the development of such unusual content? How do you explain to them that implementing in a computer game the messiness of human interactions (including conversations) will have a reasonable return on investment? How do you persuade investors to back novel, untested ideas for how characters in games can interact in ways that work as both game mechanics and meaningful human connections?

This difficulty is the reason why I support those who try even if I don’t agree with some specific approach. I’m still looking for something better than conversation trees and QTEs, though. :smiley:

how do you justify building a game with content that players may never experience?

I’ve been wondering the same about the magic system here. At least with my first investigations, it’s fantastic. Exactly what I hoped for from Underworld.

And yet… exactly as you say, so far as I can tell, this is content that most players will never experience. It’s easier and more efficient to just burn something alive or smash it in the face with a sword, than to figure out whether your limited subset of the runes will help you do something creative against them. And if you kill them they’ll just be back the next time anyway. Face smash!

You can spend over a hundred hours in the game, as I have, and complete it, and still be four targeting runes short, and still not unlock the skill that gives you max mana, so you can’t cast level 5 spells, let alone level 6!

So for almost everyone, this fantastic magic system will be just “heal person”, plus maybe a couple of wands. And that’s a tragic waste of developer time.

And random placement of monsters and items: in the quest levels, it’s needed because they’re re-entrant and reset, and you don’t want the same thing from a chest every time. But it SUCKS! Because there’s stuff that never drops.

In the Underworlds, everything was hand placed, so you could guarantee that a careful completionist player who knew the game could experience every single thing you’d created, every item type (did you know that there were two types of red ball in UW2? Identical, except one bounced on water, one sank), every monster, location, secret, spell and conversation option. They’d go WAY out their way to get that one rune that’s not found anywhere else.

But since UWA uses re-entrant, resetting areas, the mobs and loot needs to be randomized. And other than memora-collection (non-randomly placed items!), there’s no point in going out of your way or exploring the levels rather than just running straight to your goal, because you’ll be sent in there again in another quest anyway. Which means there are item types and locations that most people are never seeing.


I would add menus are not where the action happens. 2D menus lack the interactivity and depth of a 3D experience.