We should have a big ugly fight about this now and get the thread shut down over berry partisanship.
Yes, totally. And there are choice-based games that nail the tone-- the Thalia games have pitch-perfect tone and avoid the “know everything beforehand” trap by asking you to essentially pick a tone. And I agree that Ludwig did it very well in a parser game.
It’s something I’d like to see more people experiment with.
You’re too kind! And to be fair, “picking a tone” is the entire mechanic around which the dialog puzzles revolve so I would hope we got that 50% of the game right!
I think there’s parts of the Lady Thalia games that, in isolation, could be done as well or better in a traditional parser with a world model, but I think the blending we do with dialog wouldn’t work nearly as well. Working in Twine also allows us a lot more control over the player experience when it comes to tight-paced setpieces that are still (minimally) interactive - I don’t think the “escape gauntlet” type sections we added in LT2 & 3 would be nearly as immersive if you still had only two options at every point but had the ability to type something else that would throw an error. Some people may disagree, and that’s fine! But the kind of game we’re interested in making there isn’t something that I think would be straightforwardly improved by switching to a limited parser model, so they can play something more to their taste and we’ll both be happy.
If mimesis is the only thing you’re interested in then that’s fine and valid! That is, in fact, an unpopular opinion and therefore perfectly appropriate for the thread. But this is a small and tight knit community and I think not couching those opinions appropriately comes off as attacking individual people’s work, which is not typically received well. I can say all the harsh things I want here about, say, Metal Gear Solid without fear that Hideo Kojima will see it. But Ben Jackson is a member of this community and a pretty cool dude, so saying stuff like this comes off a lot harsher. Adding a strategically placed “I think” to the above quote would go a long way to avoid giving the impression that you are personally insulting people’s hard work.
I don’t think it’s reasonable to read a claim about what would improve a work as an insult to the author of the work. The Den is still a fine game that I enjoyed a lot and it totally deserved its high standing in the comp!
Don’t worry, no offence taken!
My personal hot take is that I generally prefer strawberries to raspberries, but a really tasty raspberry will still beat an average strawberry every time.
That’s fair enough I suppose, but that was just one example – as someone who works entirely in choice systems I don’t like it when people say that they’re always inferior to parsers because of xyz reasons. There’s also a long history of arguments about choice vs parser here that EJ alluded to and they’ve gotten very nasty, so people are already primed to see statements like this as an insult if you’re not taking steps to make it clear. You’re free to take this advice or not, but you might find people get their backs up less in response to your posts going forward if you do.
Raspberries taste better but those seeds, man.
Of course, we’re posting hot takes and unpopular opinions here. So maybe the etiquette is a little more relaxed here?
Honestly, I don’t think it would have been. Sure, the navigation is very parser-like, and the terminals have their own interface, but beyond that, what verbs would you be able to use more than once or twice—often enough to actually understand the system?
Yes, but it’s a difficult line to walk when this is a small community and you’re talking about people’s individual passion projects. Like yes, if you say “I don’t like x element in games” people who put X in all their games are gonna have to deal, but I’m putting my two cents in on what I think will keep this thread fun instead of descending into arguments and getting locked. But hey, if I’m wrong here then that’s actually great!
Back on topic, my unpopular IF opinion is that if given a choice between an AFGNCAAP, a configurable protagonist, and a pre-defined player character I’ll pick the pre-defined one every time!
Yeah, I do that too. Greater separation between player and character and it allows me to RP choices a little more.
I find configurable protagonists to be an annoying gimmick and when a game begins that way I always button-mash through it as quickly as I can to get it over with. I will always prefer an AFGNCAAP over dealing with that. I generally agree with you that strongly-characterized protagonists are better and more interesting, though of course it varies by genre. In an unapologetic puzzlefest, some characterization is usually preferable to none, but it should be just enough to set up the excuse plot; I don’t need an elaborate character arc. But if your game is striving to be lit-ra-shur, then a good protagonist is of course an essential part of that.
I’ll argue one strong exception to “more characterization is better”: games which pose ethical quandaries for the player to navigate. This is a really difficult thing to pull off in general; most attempts wind up just making me angry at the author and I know I’m not alone in that. The most common failure mode of such games is to pose a false dilemma between bad choices when there are obvious better options that the game doesn’t implement. Being given a player character whose backstory includes a lot of baggage is a special case of this. Should my character confess his past crimes to prevent their harm from spreading, or keep silent to preserve his freedom? No thanks: I don’t want to play that game if those crimes are something I could never imagine myself doing in the first place. I love a well-written redemption arc, but save those for NPCs or for non-interactive media.
16 posts were split to a new topic: Using Ink [split on request]
4 posts were merged into an existing topic: Using Ink [split on request]
A post was merged into an existing topic: Using Ink [split on request]
A post was merged into an existing topic: Using Ink [split on request]
A post was merged into an existing topic: Using Ink [split on request]
I haven’t properly read this whole channel, but coming in with authoring system unpopular opinions, I think ZIL is extremely well designed if you’re trying to do something extremely dangerous to the parser. (I’m joking with “dangerous”. But still, I think ZIL is great.)
There. I’ve said it. My unpopular opinion is out there.
The future of AI in IF is neither writing your game with ChatGPT (nobody wants to read something you didn’t want to write); nor using an AI intermediary layer to translate player commands (a cool gimmick, but IMO ultimately causes more problems than it solves by confusing the player about what they can actually do); nor free-form adventures hallucinated entirely by the AI (like AI Dungeon - a fun diversion, but too directionless to carry much artistic intent). Rather, the revolution will be figuring out how to get generative AI to rigidly follow authorial intent, hallucinating dynamic action and dialogue to express a world, plot, characters, and writing segments designed in detail by human authors.
I’ve done enough experimentation with this to be convinced that it’s possible, although the current generation of SOTA LLMs are too expensive to run a game like this economically. But price per unit intelligence is dropping every day, so this is rapidly becoming more feasible. LLMs particularly excel at natural language conversation - if you’ve ever played around with Character AI, imagine your NPCs interacting with that level of understanding and intelligent agency, but exactly following your intended plot and world model instead of just engaging in a random-walk roleplay hallucination. This is the future.
(At the exact same time, the future is also 100kb Z3 games that run on a Commodore 64. I said what I said.)
I think we’ll see this sort of thing pioneered in big-budget open world RPGs before it comes to IF — and the first game company to pull it off successfully is going to see its market cap exceed the rest of the gaming industry combined.
I’m not so sure. There are already advanced NPC AI systems and mods that let you have chatbot conversations with NPCs in Skyrim, but I’m talking about giving the player the freedom to do literally anything and watch the consequences within the world and story defined by the author. That’s orders of magnitude more difficult in 3D than in text, and doesn’t work at all on a game console with no keyboard.
I do expect to see more LLM-enhanced conversation trees, but the field is wide open for a smaller indie studio to build a platform for LLM-delivered authored experiences.
I’d rather see IF in pure audio book format controlled by speech. Shouldn’t be that hard(?) - still the player should say e.g. OPEN BOOK NOW, GET ALL NOW etc. (Saying “NOW” could correspond to hitting enter)