Puzzle solutions that the authors didn't intend (rant)


You’re much more likely to get a well-balanced partial-failure design (that is, one in which the player can still get by even without solving all the puzzles perfectly) if the author intentionally pursues that kind of structure than you’re likely to have it fall out as a result of emergent puzzle solutions; if they’re genuinely emergent, not planned by the author, then there’s no guarantee you won’t get five alternate solutions to one puzzle and no alternates for another, say. Whereas it is entirely possible to design with the idea either that the player should have some kind of rescue opportunity after a failure, or that the game’s overall structure allows for partial scoring. There are a few IF games that are designed with the idea of partial vs. full success, rather than success/failure – Captain Verdeterre’s Plunder comes to mind, as does the much older Paint!!!.

As for emergent puzzle solutions as a goal in themselves: if you’re interested, here’s an old blog post of mine that talks about emergent puzzle solutions and links to some other discussion.


My PhD work is exploring in this same direction: namely, how to get the good parts of simulationism into a strongly narrative story world. (I.e., not remove the author’s responsibility to tell a good story, but also not step on the player’s creativity.) I’m not feeling particularly coherent today, but interested in what others have to say about Peter’s rant.

(Andrew Plotkin) #5

Somewhat, but I’m not aiming at the kind of emergent sandbox that gets held up as a (possibly hallucinatory) ideal in these discussions.

HL is about puzzles with several author-planned solutions, not puzzles with many player-invented solutions. I think I’ve got an interesting structure (which I will not spoil at this time!) but it’s not the structure of a few very-widely-applicable tools (as Counterfeit Monkey is). If you find a solution in HL that I didn’t intend, it will probably break the game.

(Peter Piers) #6

I do concede that CVP is particularly good in not making a partial success feel like failure - it’s well written that way - but partial success is not exactly what I had in mind, because that still means a failure that a completionist-minded player (like myself) will have a hard time swallowing.

Let me put in a comparison: it’s like hints or a walkthrough versus an NPC that you can ask for help at any time. This has been the subject of much discussion already, and as far as I can tell the general consensus, which I happen to agree with, is that a player will often feel that going to external hints means admitting defeat, or partial defeat, whereas an in-game hint system (Curses, Edifice and Dangerous Curves are the three most immediate examples to my mind) is so well integrated into the world that it doesn’t feel that way.

Similarly, CVP is all about little failures and little successes. The bit in Christminster I keep talking about is where you have to successfully hide an item from the villains before they search the room… and if you don’t make it, you’ll have a chance to retrieve the item later on. Maybe you’ll score some points less, but it does not affect your overall success (and makes for a slightly different story, and could potentially originate extra puzzles).

Yes, that’s the practical side of things, and the main reason why I am so against the extreme quest for emergent gameplay: I really, really want the authorial voice, because a game by Emily Short, Adam Cadre, Eric Eve, Graham Nelson, Adam Cadre, Andrew Plotkin, et al, they are all different and the product of their creator, and I would not trade that for all the emergent brilliance in the world.

My point being, these different authors will be able to come up with different solutions to similar situations, and will express themselves differently, and will know, in their own way, how to lead a player towards their desired outcome. That’s one of the accepted wisdoms of IF game design, as I understand (and support), and is less conducive to the myriad - and potentially exponentially explosive in complexity - of situations I daydreamed about in my rant. So in the end, the sort of dynamic emergent gameplay I’m talking about might still be best served in Deadline-style chunks, like The Last Express did: emphasis on character movement and motivation (your own focus on social interaction within games goes a long way towards this, Emily), rather than an elaborate world simulation…

I think I’m straying a bit. More directly to the bit I just quoted from you: I spoke about a middle ground between the oldschool-punishment-glee and the newschool-handholding (stereotypical, of course, but it’ll do for this discussion). This middle ground, if actively sought by the author as you suggest (and it would, indeed, be a conscious design choice - in a sense, Blue Lacuna and Make it Good might be modern representatives of it… and of course, Counterfeit Monkey, when it gets wilder past the midgame point), would allow for exactly this little pearl of freedom and excitement and the knowledge that my efforts will be rewarded as long as it’s possible for that to happen - failure isn’t arbitrary not definitive, unless the story ends.

(btw, I consider it a personal victory when I get Emily Short and Aaron Reed interested in a thread I start)

Kasran: yes, from Zarf’s blog and updates re Haedan Lands we can get a small taste of what it would be like to implement full physics… and in the end, almost surely, to very very very little reward. I wouldn’t like to go to that extreme… but there’s something tantalizing about the idea that different minds will have different focuses, different mind-toolsets, different takes on the situation, and will eventually be able to solve the situation by going completely different ways.

In practical terms, I suppose this would translate to the simple locked door puzzle in this fashion. Locked door requires key. So far so good. Now we change that to “locked door requires object that has the property of a key”. Multiple objects may have that property. Other objects may have part of that property, and have to be combined to create something with that whole property.

I mean, that’s the puzzle reduced to its bare bones design (where it’s definitely not as appealing). But you can apply it to that MythBusters example I gave. The property that unlocked the door was “something visible to a flying helicopter”. “Potato canon” was just as valid as “kite”. Same skeleton.

EDIT - And just by the beginning of the blog post Emily linked I can see how this is all old, old news… The very opening lines and quote speak about exactly this. It’s hard to come up with something to say that hasn’t been said before!

So if an author actively wished to pursue this route, he/she’d be thinking more in terms of properties than the actual objects themselves. And they’d still need to make it… wel… FUN!

zarf - Yeah, like I said, I don’t think it’s ideal either, not when one starts considering all the inherent complexities. You are probably working in that middle-ground I was talking about and Emily alluded to: emergent puzzle solutions are unlikely to appear by themselves, and it’ll be just as satisfying for the player and much easier for the author to have multiple author-controlled solutions.

Where the stress, of course, falls entirely on the author, who has to take X solutions to the puzzle into account, and decide to possibly disallow a few outright (personally, I rather dislike reasonably disallowing viable solutions just because they conflict with what the author intended, but that’s way, way, way better than not acknowledgind that solution at all). It’s understandable that many authors would decide not to go this arduous route… but it’s a significant improvement in the player’s experience.


Well, you have a reputation for, among other things, being minute, precise, and detailed. I’m positive that, in your particular case, everything will be as tight as it could be. Another author, approaching it differently, might have decided to leave some things looser (note I’m speaking with only the knowledge of your sporadic blog posts on HL to go by) and polish it up, creating fail-safes only when necessary. If that author’s smart, he’d be careful enough to limit the scope of what he’s using to prevent things going out of control, and it probably still wouldn’t be at all compatible with the complexity you’re putting into HL.

But both would probably amount to the same thing. I especially loved the “get to the top floor of the mansion” puzzle in Hollywood Hijinx, which is all about object manipulation, figuring out the properties of a few objects, and having an “Ah-HA!” moment. But that (counter to the main argument in my rant) is not about heavily simulated physics. There’s SOME simulation, yes, but it’s geared towards a single solution the author thought of.

I suppose it’s more that I’m ticked off by reasonable puzzle solutions not being implemented than by craving alternate ways of getting things done. Because in the end, emergent gameplay will never give me a storyline as tightly constructed and as riveting as something the author could put me through.


Emergence requires general systems that interact with each other. (This is a necessary, not a sufficient, condition.) Without generality, there are only the specific contexts the author has thought up; without interaction, one does not have the complexity necessary for unexpected solutions.

Kerkerkruip has certainly been designed to have many general systems that interact with each other, and yes, that means that unexpected solutions may come up. I’ve certainly used tactics as a player that I had not conceived of as a designer.

In general, though, I think that IF is rarely a great match for this kind of design. Story-driven IF is mostly about the particular, and general systems are often not very useful. Why on earth would you implement a whole transportation system with trains and cars and whatever … unless it takes centre stage in your game? Unless your game is somehow built around it? But most IF would not be about the system of transportation, it would be about what happens to a particular character in a particular train.

Though I strive for it in Kerkerkruip, I’m very sceptical about emergence as a goal of IF in general.

(matt w) #8

OT, but:

Does the gameplay change when you get inside the college? Because I was recently inspired to start it by the IFDB reviews and the puzzle for getting inside the college filled me with rage. Massive spoilers under the tag.

Finding the loose cobble and figuring that you needed to throw it at the bird to get a feather to tickle the don wasn’t bad. But then… a puzzle-critical object that is in your own handbag is hidden from you until you hand the handbag over to an NPC. To distract another NPC you have to do something not thoroughly obvious and use the “person, do this” syntax, which I always hate. And then… having done the obvious thing, you lose the game with no warning on a three-turn timer. And in order to avoid this you have to time your actions precisely to something that’s going on somewhere else, so that basically the only way to discover the timing yourself would be to stand at one location pressing “z” and recording the cycle, then going off to the other location and pressing “z” while counting turns on your transcript… and then the action that you have to take succeeds for reasons that are basically impossible to predict, given that up till now the main feature of the policeman has been that you can’t get away with doing stuff while he’s watching. And every time you try something and it fails, it’s about ten turns before you can try again.

Is this an exception, or is this the kind of puzzle I can continue to expect? Because several times I’ve started up some classic of the 90s and the opening act contains something that I’d never be able to solve myself without copious note-taking, trying everything on everything, trying to go directions that aren’t mentioned in the room description, timed deaths that drop you in unwinnable states that you can’t undo… (thinking specifically also of the opening of Jigsaw and a bunch of the first act of Losing Your Grip). Is it just that I’m not going to enjoy 20th-century IF without a walkthrough, or that these games haze you at the beginning? In this case I didn’t see the forgiveness that I was expecting from the reviews – in some ways it’s forgiving in that you can usually recover without making the game unwinnable, but it doesn’t seem like there are multiple solutions here.

Participate in the 2015 Interactive Fiction Top 50!
(Peter Piers) #9

Weeellllll… most IF some years ago (decades ago, I suppose) would probably just as much about the challenge of getting to the system of transportation.

I see what you mean, though. The “IF as a story” would certainly not require any of this, and your reaction is pretty much what I’d expect from the author of “The Baron”, “Fate” and even “Figaro”. :wink: Your concern simply does not lie in this area, you have a different focus. That’s part of the authorial voice thing I said I would be very, very sad to lose. My rant would mostly be concerned with games that do not focus so completely on telling the story but also take pleasure in the very act of puzzle-solving. Take Andy Phillips, for an extreme example. :slight_smile:

Matt w - I’ve seen so much negative reaction to that first puzzle… I really literally never know what to say, because I didn’t have any problems with it. I mean, yes, I lost a considerable amount of time, there was fine-tuning, there was head-scratching, but it was never so for me that I had to ragequit. All I can say is that there are some less than stellar puzzles, and I suppose that inital puzzle happens to be one of those, but if you’re ok with keeping a hint sheet close by and not be afraid to at least try to tackle the puzzles yourself at first… maybe you’ll be left with a better impression. :wink:

And indeed, Christminster is not about multiple solutions, not really. In my rant I speak about two things, and we’re focusing on alternate solutions and emergent gameplay, but an imporant part of what I mean is also “allowing the player to screw up, without it being a failure - letting them catch up later”. Christminster also has a system where by researching a lot you’ll be able to find shortcuts between places, which will prove critical just before the endgame - but if you fail to research the correct books, as I did, you can still go a slightly different way towards your goal.

(Daniel Stelzer) #10

The puzzles inside the college have so far made more sense to me. But with that one:

The idea that led me to the solution was that I wanted to frame someone else for the theft, and I knew that the magician would take anything I offered at the right time.

(matt w) #11

True enough, but I think that that illustrates what I mean about having to rely on your actions having unforeseeable consequences

since framing the magician for the theft isn’t what happens when you give him the key. (And in fact that wouldn’t be useful, since you wouldn’t be arrested but you also wouldn’t have a key.)

Slightly back on the topic of the post

I think I’d have liked to be able to hide the key under the loose cobble, although that would probably have taken too many turns to execute even if the possibility were implemented.

I promise I’ll have something more to say about the main topic! I may have mentioned that the solution for the biggest puzzle in Faithful Companion was one I didn’t intend, because the one I intended didn’t work. (That is, I was coding it and I thought “Well, maybe a clever player could bypass the puzzle by doing this,” and then I tested my solution and realized it didn’t work, and then in a panic tested that the bypass did work so I could get the game done in the time limit.) But one thing is that I’ve thought since reading Emily’s blog post, is that if maintaining solvability is an issue with emergent puzzle solutions, maybe the author who wants emergent behavior should drop the idea of the story as a linear or semilinear progress through a series of puzzles. If you can’t unlock a gate, maybe instead of finding a way around it you could go down another path. Though there are good reasons most people don’t do this.

(Daniel Stelzer) #12

I hadn’t planned to get the key, I just wanted to get rid of the Constable. Then I could presumably climb in through the window I smashed or something along those lines.

I think it’s okay to have actions with unforseen consequences be the solution, but only if the obvious result would solve the puzzle as well. The Zorkmid tree in Sorcerer is an example:

However, I don’t think the Christminster example counts, as the solution I was trying for was significantly different from what happened (and it meant I solved the next puzzle a different way).

(Marco Innocenti) #13

In my experience, I found myself with an unexpected solution in my last IFComp. Adding to this, there was a peculiar part of the game that could have ended up in creating emerging puzzles, which I decided to definitely block before birth.

More in the spoilers.

[spoiler]The final puzzle of Apocalypse involved a roaming creature which wanted to eat the PC. I forethought of 2 different solutions. A third one arose when a player (and author) found a way to exploit the ship’s geography to bypass the monster.

This was not intended and a certain alarm kept freaking out with error messages due to a deamon never being halted on puzzle completion (the game could not understand the puzzle was complete, as the beast was still alive).

In this case, I decided to add the third solution to the list of possible ones for two main reasons:

  1. It was clever, and I thought that it possibily added to the game. In fact, I was planning to add a similar solution from the beginning but was not able to code it (or, let’s be honest, to PLAN it).
  2. I owed to the bug-finder and I wanted to prize her. So I decided to let her live inside my game forevaaahhh. The game now has an additional Achievement named after that player/author.[/spoiler]

[spoiler]An entirely different matter is the rail system in Apocalypse.

It is indeed very short and I never intended it as a “real” traveling method. Just another obnoxious puzzle.
The testers struggled a lot with that certain puzzle, and the feedback opened to a series of different solutions and planning on the same section.

In this case, otherwise, what I decided to do was to completely block unintended behavior to avoid a) the game becoming too large, and b) exposing my lack of skill in programming games.

So, the ending result is in my opinion pretty dull compared to the rest of the game. A little more time following testers’ feedback would have resulted in a much more fun section of the game (which would have probably hauled my game well beyond the 2hrs limit of the Comp, but anyway…) and a wider sense of agency. Instead of a series of scripted “<insert a way of saying ‘you can’t do that’ here>” you would have found a living mechanism, free for everybody to tickle at.[/spoiler]

None of my examples say anything about emerging puzzles, but I thought I’d share this anyway. I guess is a nice point of view on how authors can suck. :slight_smile:

(Ron Newcomb) #14

Presuming that’s true, your sentence sounds the same if you replace “emergent gameplay” with “interactive fiction”, assuming the comparison’s to regular fiction. I feel IF adds a lot if the player finds they can plan a strategy, act it out, and be rewarded for it. Lock-and-key puzzles don’t generally allow that kind of complexity, so some kind of toolset + situation + complex environment setup is warranted, which is some of the stuff needed for emergence, and some of the stuff for alternate puzzle solutions as well.

And there are ways to ensure solutions exist to a game that’s intended to have some emergence: a topological sort. Current tools don’t support that though.

(Brief definition: given a series of “depends on” relations like “A depends on B”, “H depends on Y”, and so on for forty-eleventeen such rules, a topological sort puts them in order… if there is one. The programming language Prolog excelled at answering questions of this sort of problem: “is there a path from O to X?”, “How many?”, “List them for me.”)

(Andrew Plotkin) #15

I wrote that path-find-and-sort tool, because I needed it. eblong.com/zarf/plotex/

(Because it’s built for myself, it’s effectively a mini-programming-language.)

(Peter Piers) #16

Point taken. Perhaps all the examples I’ve seen and heard about and played are not good examples and better games are out there with emergent gameplay which lose nothing of the authorial voice.

I start to regret bring up the “e” word anyway. It’s got lots of connotations, and it got me talking about something bigger than I intended. I should have stuck to my original image: the author wants the player to build a potato cannon, the player goes and builds a kite. That’s not necessarily emergent gameplay, that’s just an alternate puzzle solution in an environment physic-rich enough to work.

(matt w) #17

I’m not sure what the difference is here.

(Peter Piers) #18

I’m going to stop talking. I clearly haven’t got enough of a grasp on most of the concepts I brought up and am contradicting myself and doubling back. Sorry y’all.

(matt w) #19

I didn’t want to silence you, Peter! I think part of the issue is that the terms are nebulous and could stand some clarification.

I think of emergent gameplay as effects that are not necessarily foreseen by the designer without being bugs, arising from the interaction of game systems. But I suppose there’s another connotation where basically all the gameplay is emergent in that way.

Perhaps Jamespking’s example was one where gameplay was emergent only in a loose sense – by taking a clever route through the map you could solve a problem without using the authors’ intended solutions, but it doesn’t really come from the interaction of a bunch of low-level systems beyond the beast’s navigation and the map itself.


Aren’t most unintended solutions bugs?
For example, in Wishbringer you can


which completely avoids the need to

open the can.

(Andrew Plotkin) #21

Most games don’t have the kind of mechanics in which unintended interactions can be (a) plausible in the game world (b) desired by the player © compatible with the author’s intent for the overall game design.

Thus, we are discussing games which do.


Couldn’t this also be done by creating rules? For example, if I need to get into a locked door, most games have an A + B = C result. A is the key, B is the lock, C is the open door. An emergent way of dealing with this might be X + B = C, where X is anything that is heavy enough to break the lock. It could be a rock you picked up earlier in the game, an axe you found behind the toolshed, or the key that opens the lock, or the well placed kick of the player – if they have the appropriate strength, which could also be determined by other factors, such as not being ill, or by increasing a skill, or whatever. This isn’t exactly emergent – but it could be, if you had to combine the use of the rock, which wasn’t strong enough to bust the lock, alongside a strength potion the player just drank. The combination would equal enough force to bust the lock – so a solution created by the player, based on values inherent in the rock, the lock, and the potion.