What are your IF Hot Takes and Unpopular Opinions?

General Off-Topic Discussion
443

Technically they know how sentences about leaves, throwing, and guards might be constructed in their training material, and it probably doesn’t actually know that very well, since throwing leaves to distract people is not a normal thing.

As a quick experiment I tried asking ChatGPT about the following:

You are standing in an otherwise empty park in autumn. Leaves litter the ground. A guard is blocking the path you need to take. Describe a non-violent way you could get past the guard.

and it told me a nice story about how I engaged the guard in conversation about how beautiful autumn was. I asked again about specifically using the leaf, and it had me engage the guard in conversation about the leaf which had him so distracted staring at it, that I walked past him.

I think I’ll stand by the suggestion that, for a general LLM at least, a physics engine would produce far better emergent behaviour than this :smiley:

I also tried the same scenario where you explicitly playing an IF parser game

Summary

> look
You stand in an otherwise empty park, the ground scattered with fallen leaves. The cool autumn breeze stirs them gently. Ahead of you, a guard stands at the path, blocking your way. His eyes are fixed ahead, and his posture is rigid, clearly intent on his task. A few steps to your left, a bright red maple leaf lies on the ground.

> pick up leaf
You bend down and gently pluck a brilliant red maple leaf from the ground. Its vibrant color stands out against the brown and gold of the fallen leaves around you.

> approach guard
You walk toward the guard, holding the leaf in your hand. He doesn’t notice you approaching at first, his attention focused elsewhere.

> show leaf to guard
You hold the leaf up in front of the guard. “Excuse me, I found this beautiful leaf,” you say, offering it to him. “Isn’t it amazing how vibrant the colors are this time of year?”

The guard’s eyes flicker to the leaf, and his stern expression softens slightly. He reaches out, taking the leaf from your hand with a slight nod. “Yes, it’s quite stunning,” he replies, his gaze lingering on the leaf for a moment. “I haven’t seen one like it today.”

> move past guard
As the guard examines the leaf, you step around him, careful not to draw attention. He doesn’t notice you slipping past, absorbed by the delicate pattern of the leaf in his hands.

> look
You’ve successfully passed the guard. Ahead of you, the path stretches into the park, now free of obstruction. The quiet beauty of the autumn scene invites you forward.

Summary … it really likes showing the leaf to the guard!

4 Likes
444

I think I’ll stand by the suggestion that, for a general LLM at least, a physics engine would produce far better emergent behaviour than this :smiley:

Yeah, I think I agree with you that the physics or Inform engines are generally better tools than LLMs for producing emergent behavior in the context of gameplay, especially when there’s a world model with definite goals for the player.

My point is that people want to use LLMs because, if the LLM is adapted to the purpose of game design (whether IF or non-IF), it would bypass the need for world modelling. This may or may not produce emergent behavior in the way that players and developers want or expect, but it probably would to some extent.

And, like I said, the degree of stuff that’s actually emergent vs. planned or known by developers/designers is often not immediately clear.

445

What’s his job description again?

4 Likes
446

I’m wondering why you even need to distract him…

3 Likes
447

@moderators can we move 354, 355, 359-377, 379-381, 387, 391-400 to an ai thread? and don’t include me in it. I’ll concede leaving the og “hot take” about ai

2 Likes
448

354: is about parser v choice

355: not about AI

359-377 are about Dialog, choice narratives and mimesis.

379-381 are about characterization of the PC

387 was split some time ago.

Hot Take: If anyone wants to request a detailed surgical topic split, first, please reconsider because conversations evolve. If it is necessary please provide LINKS to the messages, not just message numbers because I don’t automatically see message numbers and I have to look at the URL that changes as I scroll through a message list.

I also have to consider motivation for the topic split.

It’s one thing if a thread devolves from ZIL to talking about cats where you can say "can you split everything after this message: [link provided] into a new thread titled “cat paws”. It’s another in a “hot take” thread to say “all these AI messages need to go somewhere else” casually when AI can definitely be considered a hot take.

Remember: If you know you are responding in a manner that is off-topic, you can split the thread at your reply. In the Edit Window, change the pencil on the upper left to “reply as linked topic”

5 Likes
449

Okay, noted.

2 Likes
450

I don’t know how hot my take is, but anyways: I think IF works best as a hybrid, or more precisely IF is the medium, your game still needs actual gameplay (otherwise it’s just a story with multiple paths you can read). I like branching narratives with engaging storylines like the next person, but I think breaking it up with other mechanics can aid with pacing and tension and make the overall experience better.

7 Likes
451

My hot take:

Less text > more text. To a point.

6 Likes
452

I don’t know how hot my take is, but anyways: I think IF works best as a hybrid, or more precisely IF is the medium, your game still needs actual gameplay

At first this made me upset, but then, after thinking more about it, I kind of agree. So yes, a textbook definition of a hot take.

1 Like
453

Honestly, imo this is a solid rule for writing in general. Less is more. Editing is more about cutting out than adding in.

8 Likes
454

This take is a nice temperature to me. Real Goldilocks stuff.

A lot of unintentionally difficult parser games are that way because they don’t have a good answer to the question “what do you do in this game?” (“Solve puzzles” is not sufficient.)

8 Likes
455

I actually think “solving puzzles” qualifies as gameplay. I just don’t particularly like that gameplay, so I stayed clear of parser games for now (also I’m really bad at puzzles, judging by how bad I do at point & click adventures). I do think there’s better gameplay than puzzles though. “Sunless Sea” has the whole trading, sailing and fighting stuff, “Trigaea” as has simple fighting system when exploring and “I was a teenage exocolonist” (which I’d still consider IF) has a card game for most relevant skill challenges. That were most games with better gameplay I could come up with right now.
Puzzles make it really easy to get stuck if they’re too hard, making you try everything or consult a walkthrough, neither of which are fun. In the 3 games I mentioned, (most) of these minigames are setbacks if you fail them, but they don’t destroy the story. Puzzles can outright block all progression if you can’t get past them.

3 Likes
456

I like puzzles, and I’m fine with games that are all puzzles without even an excuse plot, and even think it kind of weird that some versions of games like Tetris, Panel de Pon/Tetris Attack/Puzzle League/whatever else that franchise has been named over the years, or Puyo Puyo have a story mode.

That said, a dungeon crawl that’s just a bunch of puzzle rooms where there’s that one puzzle that’s super hard but blocks off a choke point in the game until cleared or is just an endless stream of super hard puzzles that have to be solved in a specific order can easily cross the line from frustratingly fun to just plain frustrating, and specific to parsers, this can be especially frustrating if the solution to that one puzzle involves some obscure verb used nowhere else in the game or leaves you scratching your head even when you know the answer(the game can have esoteric logic, but the solution to puzzles should at least make sense within the context of the game, especially for puzzles on the critical path, an outlandishly out of context or fiendishly hard puzzle is more acceptable if it’s meant as a challenge for completionists or for someone who’s already beaten the game).

2 Likes
457

especially for puzzles on the critical path, an outlandishly out of context or fiendishly hard puzzle is more acceptable if it’s meant as a challenge for completionists

I agree and would go even further. I think most puzzle games should expect players to complete far fewer puzzles than they contain.

Finishing about 60-70% of puzzles at most should let you finish a game. I think this is what most modern/popular Sokoban games require you to do to move past each world, and in turn, finish the game.

I don’t see this sort of leeway in first-person puzzlers and puzzle platformers very often, and I don’t see it in IF very often.

(IF authors possibly have an excuse, since the puzzles are often tied to the narrative and involve lateral thinking. It possibly takes more time and effort to create those than it does to create pure logic, platformer, or physics puzzles.)

1 Like
458

I think I agree for the most part, but I just recently had an experience where I think the opposite would’ve been preferrable. It was an online escape room type game with a team solving co-operatively. There were a number of sub-puzzles that gave parts of a symbol-to-letters code. These were intended to be used to solve a final puzzle with a coded phrase. The team skipped over a number of the sub-puzzles in favor of wheel-of-fortune-style brute forcing the final phrase. Not as enjoyable, at least to me.

2 Likes
459

Yeah, half to two-thirds of the puzzles being required to beat the game seems like a reasonable rule of thumb… and like most rules of thumb, there are cases where violating the rule of thumb is the right choice for some value of rightt… and yeah, a case where puzzles give you clues to a final puzzle but the final puzzle can be cheesed via brute force is a candidate for tighter than usual puzzle completion requirements.

1 Like
460

I think “not all puzzles are required to complete the game” is one approach to avoiding players becoming stuck. I can see that working for Sokoban or similar mechanical puzzle games because (1) there’s no narrative linking the puzzles and (2) the player might want to go back and try the other puzzles later. But I think an IF game with a hint system (particularly gradual hints like the Invisiclues style) is meeting the same goal a different way.

6 Likes
461

But I think an IF game with a hint system (particularly gradual hints like the Invisiclues style) is meeting the same goal a different way.

Yeah, I have no issues appreciating a game with a walkthrough or Invisiclues hints, though it does have a different feeling from games that require partial completion.

When I use a walkthrough, I’m imagining what it would be like to be good at puzzles. When I partially complete a game that doesn’t require 100%, I’m talking myself into believing I’m good enough to finish the game, possibly incorrectly.

Sometimes I get both feelings from the same game. When I played The Dig, I got a kick out of imagining what went into creating the rat puzzle after furiously consulting the walkthrough. I imagined what it would be like to solve it after hours, days, or maybe weeks of effort in the 90s.

I also got the other feeling when I solved what I’ll call “puzzles of convenience,” such as the optional light bridges that allow for faster travel. I don’t think they’re mandatory, but they make it feel like you’re getting something done and engaged with the game.

2 Likes
462

Optional content is always good and makes the game world feel larger and less like a linear museum experience. Ideally the content revealed by optional puzzles should be non-plot critical but still rewarding, such as additional lore or backstory that enriches the plot or fills in backstory but is not required to understand the plot nor solve the game.

Another player courtesy: if they need to collect 8 four-leaf-clovers for a fetch-quest, provide more than 8 to find in the game world - extras will prevent the player from getting stuck searching for the “last lousy” thing.

Similarly, but also trickier - if the player needs to collect separate specific items that aren’t generic “loot” - like you need a gas can, gasoline, the tank plug, and a replacement fuse to get a generator started, that’s four items to find. Instead of hiding them individually, create six or seven potential logical places where an item might be found and when the player discovers one the first time, place the next thing they need there - the back shed could contain the gas can, or the tank they get the gas from, or the tank plug, or the fuse, either randomly or in sequence. The cellar might be a place where a quest item can be found, but if the player has already found everything beforehand in other locations, that spot would be empty.

9 Likes