Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

How Should Games Handle Branching?

Discussion in 'Game Design' started by EternalAmbiguity, May 1, 2018.

  1. EternalAmbiguity

    EternalAmbiguity

    Joined:
    Dec 27, 2014
    Posts:
    3,144
    PC Gamer on The Council's conversation system.

    (sometimes I wish I had the temerity to talk about Game Design topics in General Discussion because there'd be more activity. Oh well)

    Skippable: So this is kind of weird. I read this article and found my thoughts branching, aha, off into certain directions I want to discuss here, rather than the specifics discussed in the article itself.

    Anyway, the article talks about how The Council, a new episodic dialogue-driven adventure game a-la Life is Strange or a TellTale game, completely eschews combat for dialogue, yet still has "meaty" progression systems.

    Other topics of discussion: Marrying dialogue systems and stats (could be mentioned in your asset thread, TonyLi, with a mention of the capabilities of the Dialogue System). Game dialogue replacing combat/game-ifying game dialogue (could be mentioned in my other thread here).

    What I Want to Discuss: In the article the writer mentions that because of some of the skill choices they made, they found themselves missing opportunities. Picking a lock, or observing something. And so I start thinking about this. Obviously, since players can miss these, they can't be critical to the experience.

    So I have to wonder: how should a game handle branching? Should these different chances be places to reveal meaningful but not critical information to a player (this sounds like the copout answer, so I'll point out that a player is likely to not consider information all that meaningful if it's skipped unless it's explicit)? Should these be largely "flavor" choices where the player is by-and-large role-playing, or asserting their character's personality, with little or no variance? Or should they lead (given enough of an impetus) to completely mutually exclusive scenarios with outcomes that make sense for the choice, even if it cuts the narrative short?

    It seems like these newer "adventure" games like LiS or a TellTale game do the first one. The second one is the hallmark of an RPG (Bioware anyone?), though some do move into the first or third category. The third category I mostly see in visual novels, where you can get very abrupt bad ends if you make a dumb choice (Root Double is a great one for this).

    Conclusion: Obviously there's no one right answer. However, what do you guys think about this? If you were going to do branching in your narrative, which of these designs would you use? Or would you use a completely different one I haven't mentioned? If so, please introduce me to it.


    An Aside: This has certainly been mentioned before, but as I was thinking about this and thinking about combat, I realized that combat, or gameplay in general, is very, very rarely branching. You're usually tasked with a very straightforward action to do, and make your narrative (or any kind really apart from pure gameplay) choices in an almost gameplay-absent zone. The only game off the top of my head that did this kind of thing was Shadow the Hedgehog (where the player has three mutually exclusive gameplay objectives during each mission, and completing one puts them on a certain narrative path--very interesting stuff), and when I played it I recall finding myself almost railroaded into a certain path. I didn't feel the game intentionally made that path easier, but more that the design was not balanced as carefully as it would need to be to make all options equally viable.

    I considered passive-aggressively using all of the 5 tags available as a hint that we need more of them, but chose not to.
     
    JoeStrout likes this.
  2. JoeStrout

    JoeStrout

    Joined:
    Jan 14, 2011
    Posts:
    9,859
    I don't have much to contribute, as I rarely play and even more rarely make story-heavy games. I prefer sandbox-style games, where the player makes their own narratives. If I'm in the mood for a story, I read a book or watch a movie.

    But I find the topic interesting anyway, so I hope my inflammatory comments spur someone else to reply. :)

    And yeah, the tags around here are pretty much useless.
     
    Martin_H and EternalAmbiguity like this.
  3. TonyLi

    TonyLi

    Joined:
    Apr 10, 2012
    Posts:
    12,670
    Indeed, Disco Elysium's combat runs through the Dialogue System for Unity:
    The voice acting is really good, too.

    Given the realities of budgets and time to ship, I'd do meaningful but noncritical branches. I couldn't justify the expense of writing, blocking cutscenes, and recording voiceover for large branches that might cut off several other large branches. At most, I might do this:

    rather than:


    (From Sam Kabo Ashwell's nice taxonomy of choice-based games.)

    But even with that restriction, you can convey so much with even a single-node digression. It's like how a single offhand compliment about someone's haircut can completely change their perception of the conversation. So, rather than lots of big branches, I'd track lots of variables that can be briefly brought into the conversation.

    Those variables can also be used to customize branches on the fly. In the most primitive example, you can use one variable line of dialogue:
    • Teacher: "Your [childGender] is bright and motivated, but [childPronoun] seems to be struggling in [class]."
    rather than writing four separate branches:
    • Teacher: "Your daughter is bright and motivated, but she seems to be struggling in History."
    • Teacher: "Your daughter is bright and motivated, but she seems to be struggling in Math."
    • Teacher: "Your son is bright and motivated, but he seems to be struggling in History."
    • Teacher: "Your son is bright and motivated, but he seems to be struggling in Math."
    This might even be too primitive to accurately convey the idea, but you can imagine how this could be expanded with a little more sophistication.

    Good stories have ambiguity -- the intentional kind, not the sloppy kind of ambiguity that comes from bad writing. They leave room for interpretation and speculation where the reader gets to fill in the gaps. This should apply to story-heavy games, too. This lets the player make their own narrative to some degree even in a completely story-driven game. What do you think? Or can it never match the emergent storytelling of KSP, or the depth of writing of a good book?
     
    TeagansDad, Akshara, Martin_H and 2 others like this.
  4. Hyblademin

    Hyblademin

    Joined:
    Oct 14, 2013
    Posts:
    725
    I strongly prefer variation in the development of the narrative leading to a common ending over wildly branching stories. Also, I don't enjoy "multiple endings".

    What I mean is that actions performed or not performed that lead to significantly distinct epilogues are no good in my book. I prefer a common ending made flexible in a way that it differs for each player based on previous choices, and different, interesting ways of arriving at the ending that carry different implications. Ideally, the player choices aren't always so obvious (>KILL HIM; >SPARE HIM), otherwise it feels formulaic and awkward.

    A downside to this is that it's less apparent that there is any effect on the progression unless the players share stories, but there are ways to ameliorate this, like tipping players off to the variable nature of the game somewhere. Another downside is that seeing every story branch can be more difficult unless clues are given to the player for how to reach them.
     
    TonyLi and EternalAmbiguity like this.
  5. EternalAmbiguity

    EternalAmbiguity

    Joined:
    Dec 27, 2014
    Posts:
    3,144
    I think I'd probably trend towards branching with convergence myself, though less because of budgets and more because I want greater control over the story. What I'm working on right now fits mostly into that framework, with main story beats and some player choice determining different interactions along a "beat." There's a convergence point near the end, which subsequently branches off into (just a couple) different endings based on the player's choices, so I guess it has elements of the mutually exclusive branching. But my goal with that is to disguise the final branching or make it not obvious that there are other options, so it's essentially the branching with convergence model.

    Another thing I plan to work on in the future is largely linear, an RPG where the player has "side quests" they can complete. If the player completes all of these a certain way then at the very end it branches off into an alternate ending, but again I want to keep this hidden so for all intents and purposes it's just linear with no branching (or technically the "flavor text" branching model, since the player can complete certain quests different ways, with the quest still ending at the same state regardless).

    Edit: I suppose the above is partially untrue. There is an element where the player's actions do matter, and they can wind up in a totally different place. During the game they court a single character, and they can wind up going nowhere with that or marrying and having children. And while it takes place tangential to that game's plot it's very important to the plot of that set of games, so it should be included. So largely "flavor text" branching, with a mutually exclusive branching side element.

    I suppose the primary concern, apart from time/budget constraints, is the amount of control one wants to have over the story, and additionally how confident one feels in one's ability to write plausible scenarios. Regarding the latter: the "flavor text" model probably requires the least amount of writing ability, in that since no choices are really "different" you're only writing one character, only building one scenario. The mutually exclusive branching model is probably next, in that you've got all these different scenarios that need to emanate plausibly from a single starting point. And the branching with convergence model is probably the most difficult, in that you need to incorporate these different scenarios and variations of a character, yet make them all plausible within each convergence.

    That's probably a part of why you get situations like Shepard, where people complain that he's largely the same character whatever you do and there's no meaningful difference between his options. And why you get complaints like the statement that nowadays, you don't get "yes" and "no" conversation responses, but "angry yes," "happy yes," and "maybe (yes)." To be clear I'm not in complete agreement with these criticisms, but they fit with my presumed levels of writing difficulty.

    An Aside: The more I think about this, the more I realize that it has indeed been done in other games, it just wasn't as obvious (and not necessarily as meaningful). Another example would be in Mass Effect 1, where before fighting the Thorian you fight through a couple of camps of "Thorian creepers," or people who've been turned into proto-husks. Man, I need to play the trilogy again. Anyway, interspersed among these creepers (who have clearly designed "monstrous" models) are a few colonists who're "enthralled" by the Thorian--their minds are not their own, but they're still human (and they still have human models). Before beginning this section you're given a number of grenades you can use to incapacitate these people rather than kill them. After this section the people you saved are alive and a number of them thank you.

    When I played I found it quite difficult to keep track of this secondary objective in addition to the standard gameplay experience. I also find it significant that the difference is largely "flavor text" (after you've done the mission, these characters stand around and don't do anything, and you don't interact with them anymore after a short post-mission "thank you"). Certainly less meaningful than Shadow the Hedgehog's implementation, but no less fiddly on the gameplay side. Seems like it requires careful balancing. Of course, the fact that both of these use action gameplay systems almost certainly makes it more difficult. Other turn-based systems might make this easier.
     
    Last edited: May 3, 2018
    TonyLi likes this.
  6. Hyblademin

    Hyblademin

    Joined:
    Oct 14, 2013
    Posts:
    725
    I don't think I can fully agree. Many instances of this sort of plot structure end up being some of the most important moments narrative-wise in various projects.

    Of course it depends on what we mean by "required"-- flavor can be implemented easily even when no regard is given to its implications against the rest of the story, and sometimes it doesn't take away from the experience at all even when this strategy is used liberally in a project. However, there is a huge potential for impact in these moments, since the perception of good character depth relies almost entirely on them. Writing interactions between characters and events which are tangential to primary plot(s) and yet can be used to firmly define the characters' place in the big picture takes careful consideration and respectable writing skill.

    That said, I think I know what you mean, in that the minimum is easy to implement without really doing damage to the project as a whole.
     
  7. EternalAmbiguity

    EternalAmbiguity

    Joined:
    Dec 27, 2014
    Posts:
    3,144
    Yeah, I'm mainly talking about the background work. When planning out how the scenario will progress, it takes less skill to just write one scenario that continues forward regardless of these kinds of choices. It takes more skill to write realistic branches off of that that spiral into their own independent scenarios. It takes even more skill to create realistic branches, yet bring them to a convergence point that makes sense.

    Each type certainly has the potential for added poignancy in the little details, stuff like specific dialogue, but I'm only talking about the minimum here, the broad scope.
     
  8. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    Personally I don't believe in branch at all, I tend to see story as single action on a "global state", that "unit story", what we call story is actually multiple unit story. Other can cal them arcs, scene or whatever. The thing is that they are state transition, not just of the world, but also the audience perception (mental model).

    Why I think it's better, well you start to see story as function (more precisely "events" in a procedural way), and plot as data, and you can see how they combine to be dynamic without necessarily invoking more works. Think about it, the same scene where a character confess love, don't have the same impact at the beginning or at the end of a story, because the state would be different and lead to different outcomes.

    It seems that it would lead to less control, in fact I don't think so, because "state change" velocity impose an inertia that structure the story. Branching is basically statically linking events instead of logically.

    However I think my description is a bit dry and use to much technical metaphor, I'm not sure it make sense to a classical view on story.
     
  9. EternalAmbiguity

    EternalAmbiguity

    Joined:
    Dec 27, 2014
    Posts:
    3,144
    All of that seems beyond me :p