What are your IF Hot Takes and Unpopular Opinions?

General Off-Topic Discussion
352

Building off that last one:

To me, the greatest strength of a parser game—any parser system, limited or not—is its predictability. The best parsers have a consistent world model, or the illusion of a consistent world model, which you can understand through experimentation, and this then lets you predict the likely results of your actions before you do them.

(Some systems, like the primordial Wander engine or the Scott Adams one, don’t actually let you build a consistent world model—every individual command is special-cased. But these games still try to give the illusion of a consistent world model; Adams’s Savage Island I has separate code for every individual object to make it sink to the bottom of the lake if you try to swim with it!)

The greatest strength of a choice-based system over a parser, meanwhile, is the ability to adapt the choices presented based on the circumstances. If you want ACCUSE COLONEL MUSTARD OF MURDER to be a part of a parser game, well, the standard way to do it would be to make a consistent system of accusing people of crimes, and then teaching it to the player somehow before it happens. In a choice-based game, you can only reveal that choice at the dramatically appropriate moment.

And this is why I think choice-based systems have a huge leg up over parser systems when it comes to dialogue, which is an inherently unpredictable thing, where the options available change based on the circumstances! Whether it’s a topic system or a menu of choices, most parser games with dialogue effectively switch into a choice-based paradigm for the conversations. (A truly parser-based dialogue system, in my opinion, would be something where you can only GIVE or SHOW objects in the world, which forces it to integrate with the world model instead of being a separate subsystem.)

(Of course, there are tons of edge cases where my nice, neat categorization falls apart. Familiar Problems fits my “parser” model, where it’s all about mastering the permutations of a system, but if you removed the text input field it would be a choice-based game. Which gets back to the isomorphism between choice games and limited parser games Daniel Franke brought up—in this case, Familiar Problems has truly absurd verbs, so I think it’s more fun to play with links than typing in DEFENESTRATE and SOLILOQUIZE and VIVIFY every time, but mechanically there’s no reason they couldn’t all be mapped to single keystrokes like in Vambrace of Destiny with the puzzles being exactly the same.)

3 Likes
353

I agree and I’d throw drag and drop interfaces in there as well. In all cases, the player is issuing commands and the system is generating responses to those commands. Really, an IF story is an operating system and arguing about the merits of choice vs. parser is like arguing about the merits of a windows-based O/S vs. a command line interface. Both can get the job done and you may have a personal bias toward one or the other.

2 Likes
354

No, it’s never been exclusively about “choice-based IF is easier to make so authors who don’t make parser games are lazy.” “Choice-based IF is inherently less immersive and less interesting than parser IF” is a take that has existed for a long time too, and “any choice-based game could be made into a limited parser game and would be more immersive and more interesting that way” is fundamentally the same argument.

5 Likes
355

Actually, I should respond to this directly, because I somewhat agree about the isomorphism but I don’t agree about the limited parser being superior. Would you be willing to play Familiar Problems for like five minutes (at least until you gain a fourth verb, beyond CONSUME, APPRAISE, and NAVIGATE)?

The version I linked lets you play with a parser and/or hyperlinks, mixed as you like. But you can imagine two hypothetical variations of this. One without the hyperlinks—the verbs are only shown when you first consume the familiar, and under a VERBS command—and one without the parser, where you can only click the links. (I can compile some non-hypothetical versions for this if we want, but that seems like overkill.)

I claim that both of those hypothetical versions would be worse than the actual version, but the link version would be superior to the parser-only version. A lot of the humor of the game comes from the bizarre verb set, but typing NAVIGATE EAST every time you want to move and APPRAISE ANNOUNCEMENTS every time you want to examine would get tedious fast. And changing it to GO and X would remove the fun of the verbs in favor of bland repetition.

(And of course, many, perhaps most choice-based games would just be awful under this isomorphism because the verbs you’d get wouldn’t be at all consistent. While I’m on this shameless self-promotion kick, I think Arkuwar would make a terrible limited-parser game because the majority of verbs would be used only once. At which point, why make players do all that typing?)

2 Likes
356

I’ll play that, but my immediate thought before having done so is, have you played A Very Strong Gland? It’s a limited parser game with an autocomplete function, such that you only need to enter the first letter of a word to have the rest of it automatically filled into the prompt. This seems adequate to address “I want goofy verbs, but I don’t want to make the player type them over and over”.

2 Likes
357

What I don’t like about choice based games is that they require you to read text about the choices you don’t take. This does break emersion. Choice prompts should be as brief as possible.

2 Likes
358

Waiting for the inevitable anti-mimesis comp/jam/manifesto.

Except the state of available tooling, the ways in which the nature of interaction, and so on all alter the degree to which the postulated isomorphism is desirable or practical. It doesn’t matter if there’s a hypothetical isomorphism if it cannot, in fact, be realised.

You could hypothetically present a game with a set-of-choices-rotated-through-before-being-confirmed mechanic like Devotionalia as a limited parser interface, but this would either have to be replaced with a different mechanic (thus breaking the isomorphism) or would probably violate your ostensible reason for preferring limited-parser games by incorporating "a strict enumeration of [the player’s options]”.

For some works, mimesis may be either meaningless or harmed by a hypothetical limited-parser version.

To reuse and clarify an earlier example, I would think Digital: A Love Story was less immersive and mimetic if I had to interact with the narrative frame of a computer system GUI through the lens of…*checks notes*…a limited parser.

For another example, I don’t think mimesis and immersion per se apply to something like Social Democracy, so it’s difficult to see how a limited parser could add much in the first place.

3 Likes
359

A Very Strong Gland was written in Inform 7, while Familiar Problems was written in Dialog. The feature may not be able to be replicated on Dialog.

1 Like
360

I am my own anti-mimesis jam

13 Likes
361

Yeah, sadly it’s a Glulx-only feature, so Dialog doesn’t support it. But that is definitely a neat way to do it!

The problem is, it requires all verbs and nouns to have unique first letters, which puts really strict constraints on how you build your game. Much as I love DiBianca’s games, I feel confident in saying Familiar Problems has much stronger flavor than A Very Strong Gland, because it didn’t have those constraints on its writing.

(We did actually try to make each verb start with a different letter at one point, for mnemonic reasons rather than interface reasons, but both CACHE and CONSUME were so flavorful we didn’t want to sacrifice one of them.)

3 Likes
362

I’ll agree with the first half of your claim here (any limited parser game could more-or-less mechanically be converted to a choice interface). But I don’t see the vice versa at all. Here are some choice-based games I’ve played recently (I’m including quite a few to increase the odds that you’ve played at least one). I can’t picture what the hypothetical limited parser interface would look like for any of them. Is there any chance you could give an example?

Bogeyman
KING OF XANADU
Civil Service
ROD MCSCHLONG GETS PUNCHED IN THE DONG
The Saltcast Adventure
LATEX, LEATHER, LIPSTICK, LOVE, LUST

4 Likes
363

This might be my bias as a long time Linux user who does pretty much everything that doesn’t involve a web browser from the command line, but tab completion or something similar should definitely be standard in any modern parser game. And while no shared first letters is a nice to have, typing ca for cache versus co for consume strikes me as a reasonable trade off of losing a bit of convenience if it improves flavor or avoids pulling in obscure or out of place words to prevent collisions… now, if there are a dozen ca verbs, another dozen con verbs, and a third dozen of c verbs with other miscellaneous second letters, trying to tab complete on those might be more annoying than just typing things out exactly… Of course, if you have that many c verbs, you’re probably either going for making the game as obtuse as possible(say, the player is cursed so they can only say words that start with c, so all valid commands for the game are c words) or being incredibly ambitious with hundreds of verbs in the game or having thesaurused every verb to hell and back.

Though, other convenience features to cut down on typing in parsers could include things like keyboard shortcuts for common actions(maybe alt+i opens an inventory menu with menu items being inventory items and you can press d to drop the selected item, u to use the selected item, g to give it to a NPC if you are engaged in conversation, s to show it, etc.) or user defined scripts that when invoked repeat a list of commands.

Of course, how well any of these could be implemented within existing frameworks, via extension to existing frameworks, or would require building new frameworks is the real question and one I’m not qualified to answer.

7 Likes
364

Have to be careful with tab completion though, or you’ve just given the player an efficient strategy to lawnmower the verb list. Perhaps only complete verbs the player has previously typed in full?

5 Likes
365

I had a rough algorithm for this written down somewhere but maybe I can remember a partial summary:

  1. Objects, by default, are not eligible for tab completion.
  2. Any object mentioned in printed text becomes eligible: room descriptions, examinations, action responses, etc.
  3. Objects not in scope are not eligible.
  4. A previously-eligible object removed from scope is eligible when it’s returned to scope.
  5. The world model reserves the right to arbitrarily set or unset tab completion eligibility for an object.
  6. Fake objects which exist only to alert the player of the disappearance of another object (an “Unthing” in TADS Adv3Lite) are eligible by default, and become ineligible after the first interaction.
  7. Verbs should be eligible by default because the problem of figuring out what verbs are supported in the game is (in my opinion) not fun lol. If necessary, a verb can begin ineligible and become eligible at some point in the game progression.
8 Likes
366

I wonder though. Everybody went on an “emergent narrative” kick several years ago, discovered roguelikes, and decided losing is fun. The chaotic behavior of a consistent world model created unpredictable situations and demonstrated that you don’t need to be fair and winnable all the time to have a good game. Somehow that “roll the dice and see what happens” attitude didn’t penetrate to dialogue systems, but in principle open-ended conversation could be fun, in a game where softlocks of some kind don’t ruin the experience.

5 Likes
367

Yes, but I’ve never liked the options I get in menu-based or choice-based conversation systems. There are many, many times where it grates on me because I don’t feel like any of them are what I’d want to ask. Or I don’t like the tone. Or I just don’t like feeling confined, and knowing exactly what all my options are. Good conversation should be predictable-- you know what you should ask. But it should also be a little flexible-- where you can ask about a lot of reasonable things. I’ve been playing around with conversation a lot in parser games, and it’s really hard to do it well, and to adapt it to what the player can know at a given moment, and it requires typing (although I like my A/T for ask/tell system OK). It requires a huge amount of testing to know what people will ask. But I feel like parsers, if done right (which is huge if) could have the leg up there. But I generally like everything about parsers better. I also like raspberries better than strawberries, which doesn’t make them objectively better.

7 Likes
368

@AmandaB
A while ago, when thinking about how dialog could play out for the user, I thought about choosing their current disposition, then seeing appropriate responses based on that. It would mean a lot of dialog writing, but if the game was focused on that aspect, why not?

I’ve found that my desire for wanting different responses (than what was available) was based on on how I felt about the situation or the character I was talking to. It’s all about nuance in most cases and can still lead down the same dialog tree (basically), but maybe your character is more agitated or reluctant. Anyway, what you mentioned struck a chord with me.

1 Like
369

Okay, with some of these examples I see your point. The simplest, CYOA-like choice-based games, where the game is just a directed graph of static pages and there’s no game state beyond what page you’re currently on, don’t necessarily map pleasantly onto limited parsers because there may not be any regularity to the types of actions that the links represent. But, games that have even a little bit of a world model are much more likely to convert well. The Den, for example, could easily have been written as a limited parser and would have been better for it. Some of your examples would be fine too, I think. I only played a few minutes into Bogeyman, but the link actions generally seem to have obvious correspondences to standard parser verbs so it should be straightforward to convert.

370

I’ve played through Familiar Problems now and I have to say I’m very unpersuaded by this example. I played with the hyperlinks just long enough to see how they work and then switched to using the parser the rest of the way through, and just found the hyperlink box to be an obtrusive blob taking up screen real estate.

The bizarre verbs are amusing, but they’re amusing once each. Once I learned “appraise” and “navigate”, I didn’t get any additional humor value out of seeing them listed out above every prompt. I played the game by using standard parser synonyms like “X” and “E”, which I was glad to see were all implemented. So, nothing would be lost for me if the link box were replaced by a “POWERS” command that provided a reminder of what I’ve acquired so far.

On a PC, I can type “x” much more quickly than I can click a link; I can type “appraise” in about the same amount of time that it takes to click a link; and I can type “appraise announcements” much more quickly than I can click two links, because it takes extra time for my eyes to scan for where the second link has appeared after I’ve clicked the first one. The calculus would be different if I were playing on a phone or if I weren’t an expert typist, but I found the keyboard to be a more convenient input method even for words that lacked single-letter shortcuts, and for those that did have shortcuts it wasn’t a close call.

1 Like
371

I agree with all this (aside from the raspberries). Short of having an LLM integrated (a la AI Dungeon), IF dialog is always going to be somewhat unsatisfying, and I find simple topic-based systems preferable to anything else achievable without AI. The best dialog system I’ve played recently is the one in Dr. Ludwig and the Devil, which avoids guess-the-noun frustration by giving you a list of available conversation topics, but leaving additional “easter egg” topics that aren’t on the list but also aren’t necessary for completing the game.

I share your annoyance when choices come out with the wrong tone and I hear this complaint a lot. But it depends on what sort of relationship you want to the player to have with the player-character. If the PC is meant to be an AFGNCAAP, then it’s best if they never explicitly speak at all and NPCs just respond to what was implicitly said based on the player’s command. But for strongly-characterized PCs, I stop minding tone issues so much because it’s clear that it’s the PC saying it, rather than having the words put into my mouth. Varicella would be way less fun if the PC were mute.