Writing Starsector

The next update will add strong narrative RPG elements to Starsector, among other things.

I feel no small amount of trepidation because this is both a change and it is a particular story about particular characters in a way the pure sandbox certainly isn’t. This necessarily constrains your – the player’s – experience of the game-fantasy and the meta-game fantasy of an “unfinished game” which has the potential to become everyone’s dreams in a free-floating quantum state… until you see it for real and it turns out it isn’t quite what you dreamed.

I suppose this seems like an awfully negative way to start off; this is what I mean about trepidation. And I am legitimately excited about sharing more of the world of Starsector, letting players dive in a bit closer and get a feel for what it’s like for people that live in this world. Find out what they think, find out a bit more about why movers and shakers move like they do. If I may say so, I think we’ve done some pretty good work!

The written wordcount has already exceeded the minimum definition for a novel (50k) a few times over by now. I’ve attempted Nanowrimo a few times in the past and always choked almost instantly. My experience writing Starsector has been a stark contrast – the words just flow! It seems so obvious, most of the time, what comes next, what feels right to be said. I suspect part of it is the constraint of the medium focusing creativity, but it may also perhaps be the very clear connection to an audience (that’s y’all out there!). A novel feels a bit like a bunch of words floating out  in (ha) space. A game, however, has a player. They must actively engage and progress. I know a player is committed in a way a reader isn’t. (Which probably isn’t at all true; people read books, after all. I’ve even read one or two in my day.)

Whatever it is, maybe I can’t rationalize it. But something works here for me in a way that hasn’t elsewhere. I’ll take it.

Let’s get to the nuts and bolts of this.

We’ve had to deal with certain constraints and design problems while adding written content to Starsector. Some of these are faced by all games which use writing, some are particular to the context of Starsector. I am not going to talk about any specific narrative beats or plot details, but I will talk about how the narrative is structured, so from a certain point of view one could derive meta-spoilers from this blog post. I think the most pure and magical way to experience Starsector would be with no foreknowledge of any of this, so I’ll give you fair warning now: if you don’t want to know anything, stop reading.


Scared ’em away? Good. Now, let me tell you all the secrets…

Let me use a question asked by TheDTYP in the forum as a starting point:

… would anyone be able to comment on just how much agency the player has in these upcoming missions? Are they generally going to be straightforward like the Red Planet mission, in which you follow the breadcrumb trail and overcome the challenges for a reward while learning about the game world, or will they be more dynamic? As in, can certain elements of the mission or the outcomes be influenced by the players in some way? Like, “I stole this McGuffin, do I give it to the Hegemony or the League to turn in the mission?” kind of stuff?

There are a lot of questions to unpack here. I think I need to start with some groundwork.

Narrative player agency is of course an illusion. A successful game narrative creates the illusion of narrative agency through clever design, stagecraft, and the willingness of the player to believe our lies. Don’t get me wrong here: the illusion is real, and it works, and it’s good. It will also break down into its constituent parts, less than the sum of the whole you enjoyed so much, if you look too closely. I think the first few times one who possesses an inquiring mind sees through this in a media work, it’s immensely disappointing. Nowadays I find I can more consciously appreciate the illusion even if I see-what-you’re-doing-there.

Remember our spoiler warning: I am going to break some elements down into constituent parts and it will dispel the illusion if you’re not careful about managing your own experience. Let us find wisdom in these immortal words:

What do I care for your suffering? Pain, even agony, is no more than information before the senses, data fed to the computer of the mind. The lesson is simple: you have received the information, now act on it. Take control of the input and you shall become master of the output.

Chairman Sheng-ji Yang, “Essays on Mind and Matter”

Take control of the input. This is your second chance to get off the meta-spoiler train.

Leaving Space & Not Answering Questions

iD software’s Quake got into my brain at an impressionable stage of my life. I must have been 12 or so, the perfect age for that sort of thing. That game provoked my interest in everything from HP Lovecraft to electronic/industrial/metal music. Did Quake have a good story? Absolutely not. Did it have a story? Barely. Worldbuilding? Uhh… (And going back and examining the development of Quake, it turns out it’s the product of a bunch of people who were tired after the whirlwind success of Doom, and in a great deal of conflict with one another about where to take the company. Romero left, iD lost its soul. That’s all another story, however.)

Quake was a collection of vaguely related aesthetic impressions thrown into a bucket and due to said development issues never given a strong overall authorial voice – but it had something due, probably, to inertia from Doom and a weird sense of “competing dungeonmasters” in the level design. I think it’s a better game for that weird mix. I think if had iD got itself organized enough to come up with a strongly directed narrative, it’d have been super dumb. It’d have broken the magic of Quake. (See: Quake 2. Fun enough game, dumb as rocks. Arguably same deal for Doom, but I think that game works for similar reasons to Quake, though the mechanics are tighter due to more cohesive teamwork.)

Anyway, what really got 12-year-old me was looking out into those blocked-off parts of the maps where you heard that howling wind and watched as layers of purple clouds flowed over each other. What kind of world was out there? I wondered. Surely something bizarre; scifi gothic horror beyond imagination populated by the best kind of monsters: the kind you never see. What was most interesting about Quake was what wasn’t there; what was suggested. The greatest horror movies never totally answer what, why, and how. The answer will always be disappointing; the shadowy gaps are where fear lives! Embrace it!

It’s not just horror. If I tried to explain the scifi principles behind everything in Starsector, a thousand intense physics nerds would tear me apart with their fascinating equations and they’d be right and I’d be wrong. There are questions in Starsector that I shouldn’t answer, and Ludd willing, I won’t!

This is part of the illusion. If the game evokes rather than answers, then the illusion takes hold in the player’s mind. Like we offload 3d rendering into a specialized piece of hardware, it’s like I’m offloading rendering of the narrative into your brain.

Of course no one wants to read a story where every fact is hedged with a dramatic “but maaaaayybe it’s this! Or that! Or even over there! Or nothing at all.” *waves hands wildly*

Yeah, that’s insufferable.

Appealing To Players Without Disappearing Up One’s Own Jump-point

It serves a narrative well to exercise restraint. The worst thing I can see when loading up a new RPG is a huge infodump about the world and what’s going on and who I’m supposed to be and the ten thousand year history of the kingdom of whatever and hey, guess what Artemisia Sun’s favorite food is! I’m going to tell you! But I don’t care – yet. I appreciate that the author is excited about showing off their work to their audience, but that care has to be earned. Start a game by framing a player’s experience of the game-fantasy with a simple archetype. Starsector has it easy: this is a space opera, you’re Han Solo1. Figure it out. When you’re ready to start asking questions, the answers will be in the game.

I’ve gotten into playing D&D this past Year of the Plague – and more than playing, I’m also running a game. (Big surprise; right.) It’s helped me think a lot about what I find myself doing to make the game relatable to players. These are my friends, I know them, and I can come up with something I think they will find compelling. If they grab onto that hook, I go further with it and riff with them and see how their choices change the world.

I can’t exactly do that with Starsector, as much fun as it’d be to run that tabletop RPG. Computer games are a different beast. But I think some of the principles can apply, you just have to apply them more broadly.

What Alex and I try to do – and not just in narrative terms – is try to anticipate generalized player archetypes. Some people like exploring, some like building, some love fighting, some like intrigue and “lore”2, some people are named Megas and want to optimize everything. We try to imagine how each of these archetypes would react to situations we set up: uh oh, those Pathers are at it again! How can I respond? Just go down the list: fighting archetype blows them away, intrigue tries to talk them out of it, maybe someone pays them off, maybe someone else gets someone else to fight for them, the lore bore tries to wheedle some backstory out of it somehow…

This applies to more than just fleet encounters of course. If we send the player into dialog with a character and the topic of Big Things In The Sector comes up, I try to anticipate how different types of players will react, then allow them to take a stance on the issue. Say we’re messing with Domain-era artifacts, how do you respond? It doesn’t have to be tremendously complex – and it doesn’t have to be so simple as a good vs. boring vs. evil choice. Building toward archetypes let’s us focus the range of possibilities toward something manageable in the face of a potentially exponentially exploding number of possibility states.

To back up to a higher level, it is important to say here that the narrative is not solely driven by what [I think] players’ want. Media created purely and cynically in-anticipation-of or in-reaction-to viewer feedback is generally… bad (and is absolutely a certain level of Hell for any creative person).

So yes, I am telling a story with my own voice, my own weird ideas and interests3. But we are also making a conscious effort to make the story inviting to the players we anticipate will show up to experience it.

What’s this look like in the game, then.

Long-form Narrative Mission Arcs

We have missions. You’ve scanned some derelicts, surveyed some planets. Did you find the red planet? Nice! That was one of our first test-cases to see if the mechanics of campaign-level missions could work and if they’d be appealing to players.

Answered: yes, and yes. Thank you for being good test subjects.

Imagine a series of missions like those, all strung together. Let’s say we introduce recurring characters. There are random elements so you don’t quite know what to expect every time. There are different approaches possible; maybe you fight, maybe you talk, maybe you bribe, maybe you find some other way to deal with a situation. There are choices to make: some change your outlook immediately. Some don’t matter. Some don’t seem to matter until they come up again way later, and you’ll be like “Alex and David, I see what you guys did there.”

You can lie, you can con, you can betray. Maybe you’ll keep getting away with it. Maybe not.

We’re getting somewhere! It’s pretty cool, I think. And I don’t have a lot more to say about all of that than, yes, we did stuff and I can’t wait to see how people play through it.

Now let’s temper some expectations: Factions will come into play, the player will interact with prominent figures from most of the factions, but we’re not making players swear loyalty to them as part of the primary storyline. That said, having a commission or extreme reputation with a faction one way or another will impact certain situations.

Also: there will not be an ending to the game in this update.

(Not yet. But you will “determine the fate of the Sector by your actions” before this game is finished, that much is certain.)

Having the experience of implementing a series of large narrative arcs, once we see how this all plays out, we’ll be in a very good place to see where we want it to end up. And we’ll know how to get it there.

Having addressed that big picture question, let’s dive back down into some more details.

Dialog Mechanics

We’re building Starsector’s narrative largely on text. What can I say, the medium is cheap! So I hope y’all like reading, because there’s going to be a bit of that involved. (I shall also try to be considerate of people who hate reading, mind you. That’s definitely one of the anticipated player archetypes.)

So unless the text is purely info-dumping, there’s got to be an interactive component. Make a choice, get some different results. Make different choices available based on previous choices, or based on other conditions that can be drawn from other mechanical attributes within the game (fleet strength, credits held, location, faction relations, person relations, arbitrary variables set elsewhere).

What about player skills, you ask. Welllllll…  Disco Elysium brilliantly inserts RPG dice-roll mechanics into dialog, and better than skill checks (though it does those), they are so often tests of the player’s character. And rightly so, because that’s a game that’s ostensibly about solving a mystery but really an exploration of a particular character’s psyche.

Starsector isn’t quite that. Starsector must function as a game, yes, but is must function as the game it already is, which is space fleet combat-focused. So while we discussed the idea of doing skill checks in-dialog, ultimately decided that this would expand the range of mechanics without meaningfully enhancing the game experience. Imagine our design process: the player experiences a challenge via dialog. Which of the 40 different skills will solve it? How many different options do we, as developers, need to provide so that it doesn’t feel extraordinarily limiting? Does this align at all with the player archetypes we’re anticipating? What if we end up using Weapon Drills twenty times but Auxiliary Support only once in dialog challenges? The player has no way of knowing that they chose a skill useless to dialog back at level 3. (We also thought through using the sum of skills chosen in a category or aptitudes as a basis for character type. It seems compelling at first because it allows for neat categorization, but once you parse through the implications it doesn’t actually make the game more interesting.)

When you’re running a D&D game you can just make up skill checks that “happen” to align with skills you know the players have. We have no such luxury.

Yes, en masse insertion of skills into dialog could theoretically be balanced and made to work, but that’s a lot of work (and it’s absolutely not responsible to try to shove an additional huge axis of mechanics into finely tuned interdependent game systems at this stage in development). Disco Elysium made it work with a huge number of skills because that’s literally what the game is about. Here, we’d just be adding absurdly intricate icing on top of the cake. Then the icing would fall over and look terrible, probably.

Take a step back. Restate our assumptions. Dialog option restrictions per skill would lock out most options for most players most of the time for most of the game. The player-experience return on developer time invested would be terrible. Besides, what do dialog option skill restrictions provide that doesn’t already exist?

How about instead we assume the player’s space captain is a generally competent person. Then instead of locking out options based on skill restrictions, let us take a positive route and allow the player choose the way in which they excel at the time of the choice as it comes up naturally in the dialog. Let the player lie and send a fake cryopod to the mercenary demanding they hand over the VIP, giving the player enough time to make their escape – AND take a finders fee of 20k credits from the merc. If that’s who the player wants their player to be, that is.

We even have a system for this already: Story Points.

Conclusion: we don’t as a rule do skill checks in dialog – there’s an exception, but it’s largely for flavor – instead we use story points to let players choose when and how to be a totally awesome hero of their own story.

Speaking of that-

The Player Character’s Character

How do you possibly give players a reasonable range of choice without writing tens of options, and then allowing the compounding implications of those options to expand into hundreds, thousands of possible narrative cases?

Some games define a character for the player that’s a particular person who has their own opinions and outlook. Most games define at least a character who’s a bit of a cipher, allowing them to center their game mechanics. Master Chief or Gordon Freeman are pretty straightforward; they raise few questions about player agency outside of the limited mechanical range that the game can excel at. These are useful constraints.

We’re in a similar place with Starsector. This game has a nature to it, and it suggest a certain type of character: you’re some kind of cool space captain. Imagine a more successful Han Solo (which I guess makes you Lando, haha). I can safely assume that this character won’t have strong opinions or interest in something like farming except insofar as they’re buying, selling, or stealing cargo related to it. Good.

Still, we’ve got to keep dialog a bit vague. We have to leave an openness in interpretation of actual tone and wording so that the player can project enough of their own interpretation to those words that it feels correct. If we provide a dialog option that feels out of place or that pans out vastly contrary to the player’s expectation, then we lose narrative verisimilitude. The illusion wavers!

At the same time, it’s very important to give the player the opportunity to drop cool lines; there’s a time and place for specificity. The art of it is judging where to apply that and where to back off.

On Secret Worldbuilding

There are big conflicts and mysteries in the world of Starsector. If you’ve played the game, you know what they are and you probably have some theories. You can infer that the story has to have something to do with them because we’ve put so much effort into setting them up; it’d be absurd to throw that work away. And it’d be cheap to pull a shock turnaround that contradicts everything already established. There’s a light cone, and the range of possible endings falls somewhere within it.

The worst thing we could do is explain everything. The second worst, explain nothing. My goal with the writing of Starsector is to evoke something greater than the sum of its parts, and part of doing that successfully is knowing when to shut up.

I’ll be doing that now. Enjoy the next update!

1: Please don’t sue us, Disney.

2: I kinda hate the word and concept of “lore” as applied to videogames, though I’ll use it. It feels like a failure somehow if “lore” is a strongly separated concept from the media work itself. This probably has structural roots in media works being consciously acted upon nowadays as intellectual properties that must have profitable derivative works spun off, and… save this rant for another day. (And I say all of this as someone who played Brigador and spent all my money on lore unlocks before anything else, so don’t take me too seriously.)

3: Alex has been very supportive of running with my weird ideas, and very kind in how he tells me when they go too far. The creative freedom he’s given me is spectacular, and I want thank him for it. Thanks Alex!

Comment thread here.

Tags: , , , , , , , ,

This entry was posted on Thursday, December 17th, 2020 at 2:04 pm and is filed under Development, Lore. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.