This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Disagree on all accounts. When AI does content, it’s ultra generic, has no sense of tone or effect, and lacks any of the idiosyncratic spontaneity of even sloppily put together human content. There’s also a lack of broader complexity, meaning that any ‘character development’ it’s adding to a given story isn’t corresponding to a grander vision of what that story aspires to be or is about, but is essentially just the cut and pasted clichés, tropes, and emblems of other works that only incidentally contains some of their meaning. I’d prefer content to be cut rather than given over to AI.
At least some of the sameness of AI writing reflects misuse or unfamiliarity with the tools, rather than a deeper problem with the technology itself. It's not hard to give different flavors to different characters. It actually takes some effort to avoid going too hard on it. Completely avoiding the dread em-dash requires a bit more effort, and keeping a world consistent requires a decent (set of) lore bibles, but it's definitely possible and a bit easier than building and keeping coherent a more serious effort by human writers. The more complete your lore bible, the more the LLM can give the appearance of talking about a consistent world.
It still needs human review -- I left in a prompt for the foxman merchant version that has a logical error with a pronoun, and did regen one response for the hedge mage that had a sentence structure error giving bear tails a claw -- but that's a lot faster than manual writing even for that.
More options
Context Copy link
That's what I think about many video games! Given that RPGs were never my forte, but I have played enough games. I've even played enough of recent-ish titles.
Take any Bethesda game or probably most other open-world titles. They all have mountains of generic filler called content that doesn't get cut despite being generic filler. The content doesn't get cut, because it needs to be there. Players like wandering into an interaction and they like doing the thing. That's the appeal. Developers can reward players with do-the-thing-get-thing reward and writers reward players to do the thing because they've become invested in some story or consequence behind it. A great game rewards a player with a dopamine did-thing-got-thing and it rewards the player with an engaging story. There are not very many great games and there's only so many opportunities for great writing in a given game.
In Starfield, there's a common loop. Player meets character NPC. Player may have up to 4 distinct interactions with the character. Possibly one or two of those interactions have 2 different variations. The player is provided with a few sentences of backstory in some way, then the player is expected to recognize the shape of a familiar story and fill in the blanks. These storytelling opportunities come a few phases translated to video game format:
There's nothing in there that can't be improved upon by a writer working with an LLM. If nothing else, this results in the player being provided the opportunity to add depth to a bland and boring A-B experience. The generic shape of the story, where the player is expected to recognize it to fill in the blanks, gets more filling.
Starfield is a bad game, but Starfield had so many of these generic fetch quests, generic storylines, generic dialogues that I don't think I got close to finish it. And hey, I know this developer, I expect some level of generic human slop, but boy did it seem bad. On the other side I've also played most of Baldur's Gate 3. BG3 is a much higher quality game. As I understand it's considered a generational class of game. My generation of this genre would be The Witcher 3 which, as I recall, had relatively rich stories and writing in part. There's still plenty of bland, formulaic, or marginal content that wouldn't be harmed by curated robot slop.
Now I could not be aware of the new fangled indie RPGs true gamers play these days, but I have played enough games to know the writers phone it in no more than a good prompt. That may be due to a workload as is typical for the industry or it might be that video game writers write games for a reason other than greatness. In either case I bet there's a use case for this now. Today! Someone could go find banal interaction in a game, feed a few prompts, and get something that enriches that experience. No question in my mind. AI will not single-handedly create a cohesive BG3 story board and 100 hours of dialogue in one go, but even a free model can help a mediocre writer enrich their 15 minute mini-story side quest #121.
There's nothing in there that can't be improved as prose, but are you entirely sure that the changes will be improvements as game writing?
I like Table Top RPGs, despite them being worse than some Computer RPGs in every way but one, and the one way they're better is the way that matters here: in a TTRPG, your players don't have to be railroaded nearly so strictly. When the players try to dig deep into the interactions with some character, there can always be something rewarding they can dig deep into. Once the Game Master runs out of official quest writeup material, he can start to improvise, and those improvisations can actually affect all subsequent gameplay. It's quite common for players to develop an attachment to someone like that elderly forgotten veteran NPC, who the GM can then slot into other parts of the story, on the fly, as a recurring side character, making the story much more fun and interesting. In the longest-running game I run, my players have one originally-mid-level mook who's managed to escape enough fights to become a recurring villain (with some hilarious banter), and even have another three mooks who (via vast interleaved efforts of diplomacy and subterfuge) they've managed to semi-reform and (despite some lingering head-butting with PCs and each other) recruit as underlings. The written adventures for this campaign included some designed-as-recurring-character NPC friends and villains, too, of course, but these four were all characters who were written with at most a short backstory but who were expected to be eliminated in the first encounter if the players had been aggressive enough and their dice rolls lucky enough. We're all glad they weren't.
In a CRPG ... do you want to let the AI rewrite your game on the fly, like a GM does, not just write things you can review in advance? Writing on the fly is probably an AGI-complete problem. If you've got an LLM that you trust not to make its part of your game worse than your part then you might as well let it write your part too. But if all your writing is done in advance, that won't let you have long-term effects on the story. The possibilities you'd have to write for grow exponentially with elapsed gameplay, as more story elements arise and more combinations in which they might affect Ascended Extras' actions accumulate. If you instead do a lot of writing in advance without letting the now-fleshed-out side characters have long-term effects on the story, that just tricks the player with false affordances: instead of interacting with a world where ten characters have deep dialogue trees and obviously are critical to the story and another hundred characters quickly get to a loop with nothing new to say and are obviously scenery after that, you'd be giving them a world where ten characters have deep dialogue trees and are critical to the story while another hundred characters have deep dialogue trees but are still going to be plot dead-ends after those trees are finally exhausted.
Roger Ebert infamously took the stance that "video games can never be art", which was nonsense, but the interactivity of games is a bit of a two-edged sword: on the one hand it's an additional capacity that can make video games much better art than non-interactive media, but on the other hand it puts the artist even more at the mercy of the audience than is the case in other media. Someone may fail to understand what you intended them to understand from your painting, but at least once they're part of your painting's audience they'll see what you intended them to see. If you want to make art in the form of a game, however, everyone in your audience is also your collaborator, and your job isn't just to make them understand a finished product, it's to guide them into helping properly finish that product with you, and part of that guidance is making it easier to see which parts of the work they should focus on the most and which are just intended to be out-of-focus background. Making the background more beautiful would be an improvement, all other things being equal, but making it more beautiful without accidentally bringing it to a spot in the foreground where it shouldn't be is much trickier. The reason why new fiction writers always have to be told to be unafraid to "kill your darlings" is that it's true but non-obvious that most authors' writing can be best improved not by expanding it but by cutting it, removing the digressions and infodumps and red herrings and detached side plots and on and on until you're left only with the things that most contribute to the story. Game writers (and level designers, and so on) have a much harder problem, because even if you avoid handing the player a pointless distraction the player might seek it out anyway, and they'll enjoy the game less as a result even if they don't understand why. I recommend playing the Half Life 2 Episode 1+2 with Director's Commentary - some of the most interesting tidbits there are tricks with which they coax players into actions as simple as looking in the right direction at the right time to see a scripted event, while not actually taking any control away from the player or even letting most players realize they'd been maneuvered into making the decisions they did.
Absolutely not. I'm just pretty sure games haven't perfected process and that there is for improvement by using AI during their creation. No game developer should trust the consumer's wisdom. I would recommend they ignore gamers the most.
If it does it well because it's powered by the 200 million dollar Unreal Storytelling Engine, yeah, I'd try it out. I don't think we need procedurally generated stuff to have more choice and consequence in a game as mentioned. The best practices reasoning ignores all the limitations, constraints, industry standards, general audience, and on and on. Pressures that can be mitigated by an always on, if not always deferred to, modest writer.*
Can we not provides clues or even tell the player what they can expect from the 100 'non-essential' characters? Don't many games already do this when they choose to include such content in the experience?
I don't need to they were clearly an innovator if not a pioneer in "show don't tell" for video games. I'm not saying to take creators, artists, and innovators out of the equation or to cede your storytelling role to a robot. Guard rails are good and necessary in many cases. Based on my experience with LLM (prose, given) and my experience with video games: there's already a role for tapping LLMs and it would be an improvement in many games-- including big budget titles. But I appreciated the perspective!
More options
Context Copy link
More options
Context Copy link
Starfield was heavily panned as having by far the worst writing out of all Bethesda games. In a game like skyrim even the useless fetch quests have some variety in the framing
Starfield’s writing is no worse than that of Skyrim, Oblivion, or Fallout 3/4. Morrowind also had bad quest writing but was elevated by the the weird fiction aspects, some good worldbuilding, and narrative constraints imposed by a tight deadline that mercifully limited quest text volume.
More options
Context Copy link
More options
Context Copy link
I was thinking of adding some caveats more or less for what you’ve mentioned. Games that aspire to create extremely vast world’s for the sole purpose of escapism, games that focus more on modularity and repetition than complex storytelling, games that have some sort of unique visionary use of AI in mind.
And I could see content created by writers who make use of AI being OK as well. The artist can compensate for the AI‘s weaknesses while using it selectively as a specialized, rather than all purpose, tool. As AI does have some strengths, mainly as a search tool. It’s hugely deficient if you ask it to be creative though, and I don’t believe it’s doing much more than merely copy pasting content from its database.
In any case, I resisted this impulse on the basis that most of this stuff seems either degenerate or a suspect.
Funnily enough, I was thinking exactly of the Elder Scrolls as a series that in some cases has produced slop content of a hugely superior calibre to AI. I speak of Morrowind here. There is no way AI by itself would ever come up with exceptional banalities like three naked Nord barbarians who’ve been identically robbed by witches or dirt farmers giving you the exact same encyclopaedic digressions on regional geography anytime you ask. AI can give you banalities all right, but not of any variety that suggests an underlying meaning or humour to it. These aren’t examples of genius, but they are examples of the kind of colour and charm that humans will give you even in the service of creating slop, and in both cases, they work not just as bland content, but as stuff that enhances the actual quality of the game. AI can’t resist reverting to the generic, so it would entirely come down to human creators to invent these sorts of bizarre outputs. In fact, in many cases I think it’s the twists and turns of the creative processes themselves that humans have to go through that leads them down these unlikely avenues. Case in point, the absurdity in these examples was a direct product of what humans had to do to try and overcome their material limitations.
No it wouldn't come up with that from scratch. That sounds like the magic sauce. The intersection of absurdist yet earnest charming humor, games are meant to be fun mindset, and creative freedom. LLMs can imitate the style. As a rule I try not to share LLM outputs, because it isn't interesting, but try variations of this prompt:
I added a word here or there, including an attempt with "nihilistic" added to iconic. Only one of them was awful-awful. Claude even made me smile with one bit. It may have been a real bit, who knows? You could work on the prompt and continue to pay a comedian to work the material to end up with 5x Lazlow radio segments. Or, consider this recent viral Twitter post on output allegedly spawned from a single prompt. But I'm devolving us down to is AI good which has been done to death.
I suspect LLM's are already quite capable to assist in weak areas (as I see them) of mediocrity in video games that require text. If a human is already doing things at a surface level, like the elderly veteran trope, then AI can provide more styled meat or ideas for the meal.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link