What is moon logic?

Authoring General Design Discussions
1

I’m currently writing a game with a puzzle that suffers from moon logic. The game is finished and it’s been out for alpha testing and the testers agreed. I don’t want to change the puzzle, so how do I provide in-game clues?

I thought I’d ask ChatGPT for some ideas. I started out by asking it whether it knew what moon logic was. Quite to my surprise, it did and I thought the response may be helpful to others, so I share it here (slightly edited).

Me: Do you know what “moon logic” is?

ChatGPT: Yes — “moon logic” is a term gamers use for a very specific (and often frustrating) kind of puzzle design.

What it means

Moon logic refers to puzzle solutions that are:

  • Extremely obscure
  • Illogical or unintuitive
  • Based on leaps of reasoning that most players would never naturally make
  • Sometimes dependent on tiny, easily missed details

It’s basically when the solution feels like it came from the moon.

Where it’s common

The term is most often used for:

  • Classic point-and-click adventure games
  • Some older PC games from the 80s and 90s

For example:

  • In The Secret of Monkey Island
  • Or games like King’s Quest

Players sometimes had to:

  • Combine random inventory items with no clear hint
  • Perform actions in a specific order with little explanation
  • Think in bizarre, designer-specific logic

Example of moon logic

Imagine that you need to open a locked door.

Normal logic: Find a key.

Moon logic:

  • Use a rubber chicken
  • On a pulley
  • To distract a guard
  • Who then drops a candle
  • Which melts wax sealing a hidden keyhole
  • That only appears if you examined a painting earlier

Nothing clearly signals this chain of reasoning — that’s moon logic.

Why it happens

It usually occurs because:

  • Designers know the solution already
  • They assume players think the same way they do
  • Clue placement is too subtle (or non-existent)

Broader use

Outside gaming, people sometimes use “moon logic” jokingly to describe:

  • Overcomplicated reasoning
  • Wild conspiracy theories
  • Decisions that make no practical sense

If you’d like, I can also explain how moon logic differs from clever puzzle design — that’s a really interesting distinction.

I then went on to ask it for suggestions for in-game hints or clues that would eliminate the moon logic for my particular puzzle. It clearly “understood” the situation and gave me a well-reasoned set of steps that I could follow, together with alternatives. None of these affect the puzzle itself, only the in-game events and clues that could lead to the solution. All of a sudden, my opinion of ChatGPT has changed.

7 Likes
2

Yes, I think the LLMs are now getting surprisingly good at acting as a “creative assistant”. I’ve been using them to help my son who is autistic to aid his visualization. He sometimes struggles to get started on problems that are completely open, e.g. draw a picture of a summer day. The LLM pattern and context matching actually does a good job of coming up with different ideas and he can then complete the task as, to him, a seemingly unsolvable infinite problem has been condensed to a smaller number of options that actually act to prompt his own ideation.

Unfortunately, in the past few days my own opinion of AI has gone the other way. Whilst creativity is great, sometimes you just want a non-creative answer and AI is still not great at not making things up, even where the problem seems simple to a human. I recently asked AI to provide me with the debugging parameters for a reasonably well known software tool. What seemed like a simple task to reference the technical documentation and provide me with a single answer turned into a two day battle of wills, threats and recriminations as the AI finally accepted that it had made up a debugging parameter that did not exist and that the software could actually not do what the AI had been very plausibly providing detailed instructions for! The AI and I are still not talking after this latest incident…

3 Likes
3

This made me very curious why it’s called “Moon Logic” but, happily, I found that and, yeah, that was a terribly obscure puzzle.

4

I think ChatGPT is the only GenAI tool that I’ve used in the past and the experience hasn’t been good. It would tell me stuff that was clearly wrong, but in a very authoritative manner. I would correct it and it would apologise and give me the correct info. It would then go on to tell me more stuff that was wrong and the cycle continued. A few responses later, it would revert to the original thing that was wrong, so it clearly has no “memory”.

This time around, I was short of ideas and had nothing to lose. Maybe it’s now using a more recent model. I still think I’ll only use it as a last resort.

Anyway, to get back on topic, I thought it’s description of “moon logic” would be useful for anyone that has perhaps heard the term, but didn’t know what it meant. Of course, you could just as easily Google it, but this was just my first question before asking it about my real problem.

5

TVTropes has an extensive list of examples of moon logic in puzzle games:
https://tvtropes.org/pmwiki/pmwiki.php/Main/MoonLogicPuzzle

5 Likes
6

I think it’s either “logic based on somewhere that is not earth” essentially, or something like “it requires going to the moon to sort out”.

Portal 2 actually created a “moon logic” puzzle

[!warning]- Spoiler
In the endgame, after learning that the “white paint” which allows portals to be placed on flat surfaces is actually made of crushed moon rocks, during the final battle the player is given high motivation to fire a portal onto the surface of the actual moon which causes a vacuum through the local portal on Earth and sucks the demi-antagonist out into space. They actually did a great job of making it obvious what to do, but it’s a ludicrous and absurdly hilarious Deus-ex- Machina denouement that shouldn’t work as well as it does.

The classic quoted example which I think is on the TVTropes page was Gabriel Knight 2 (or another one?) where the player is expected to use a piece of scotch tape on a cat litter box to pick up cat hair to fashion a “mustache” that apparently is a good enough disguise to fool the authorities.

3 Likes
7

It was Gabriel Knight 3, you had to use something sticky on a fence to rip the cat hair directly off the cat, and the real kicker is that you had to make a fake mustache to impersonate someone who didn’t actually have a mustache (the biggest piece of moon logic is that you were supposed to deduce that since you didn’t look anything like the other guy, you could draw a mustache on the passport you stole and then make the fake mustache and then there’s at least one way that you’d look similar).

3 Likes
8

I hadn’t heard the term before. Glad to learn it. I think I would summarize this as any puzzle that requires such large leaps of logic that they could only be solved on the moon. “one small step for [a] man, one giant leap for mankind.”

3 Likes
9

Apropos of this thread, Jimmy Maher posted his write up of Gabriel Knight 3 today:

2 Likes
10

Regarding the Portal 2 spoiler, I don’t think it’s really accurate to call it a deus ex machina when within the logic of the story it’s a clever solution that the character thinks of under their own power. The problem with a deus ex machina is that it’s unsatisfying to have a problem be suddenly resolved by an outside force with no real agency on the part of the protagonist, but that’s not what happens in Portal 2.

4 Likes
11

I get what you’re saying. Portal 2 does set up the explanation throughout the story, but it’s still a way random sequence of events that has to happen fortuitously for that outcome at that moment. Just like a helicopter happening to fly over a marooned raft in the middle of the ocean and happening to have a rope ladder. Agree to disagree. I will say it is not technically “moon logic” in the sense we’re describing since it is streamlined not to confuse the player, but it is literally “how the hell did that even work?”

[!warning]- More peripheral/spoilery talk about Portal 2 finale
And seriously, I don’t think any player ever honestly even has time in that moment to consider what they’re doing as a puzzle and solution. The circumstances are actually set up so it will happen accidentally even if the player isn’t predicting the outcome.

The game manipulates the player to place one portal directly under Wheatley to get the portaled rockets to hit him and not be deflected by the shields, and then sprays water everywhere washing all the white gel away so there is no other suitable location for the player to move the existing portal onto, and the only other option is to shoot the moon. They even nudged the engine to make sure the player has the camera pointed at the ceiling when it collapses to reveal the moon, and stick there pointed up long enough to see the shot hit with the “speed of light” delay. I think they even said they accepted either button press (orange or blue portal - regardless of which one was actually in place on the floor) to shoot the moon portal because they didn’t want any confusion or delay in that moment for pacing.

The wodrldbuildling for it wasn’t even planned but retconned in - they came up with “shoot the moon” finale idea before the white gel was canonically made of moon rocks.

https://youtube.com/shorts/nZPzT5sF2eU?si=ilqig9ceyu1LwSaJ

2 Likes
12

I thought that moment in Portal 2 was brilliantly executed because of exactly all the things you just mentioned. But I wouldn’t call it a moon logic puzzle either - the whole point of a moon logic puzzle is that it requires completely inexplicable thinking to get to the solution, whereas what you’re talking about is a setpiece that’s barely a puzzle at all.

5 Likes
13

To me, the Portal 2 instance isn’t moon logic because after the fact you can understand why it happened. I would group puzzles (for the sake of this discussion) into four broad categories:

  • Trivial: It’s obvious beforehand what you need to do, so you do it. The grate is locked; you’re holding the keys; UNLOCK GRATE. Now it’s unlocked.
  • Satisfying: It’s not obvious beforehand what you need to do, but once you’ve figured it out, it’s obvious that that had to be the solution. A troll demands a treasure to cross his bridge; there’s a treasure that can be summoned back to its starting location with a magic word; use the magic word to scam the troll.
  • Unsatisfying: It’s not obvious beforehand what you need to do, and once you’ve figured it out, you go “that makes sense but I never would have thought of it”. A magazine has a label mentioning a specific person on it, so you carry it across the map and drop it in a room where that specific person’s name is mentioned to get a point.
  • Moon logic: It’s not obvious beforehand what you need to do, and even afterward, it doesn’t make sense why that was the solution. A ferocious dragon is threatening you; you need to KILL DRAGON, and when the game asks “with your bare hands?!”, say YES. That kills the dragon. Why was that the solution? It doesn’t make logical sense even in hindsight.

In Portal 2, I couldn’t have predicted beforehand what would happen in the finale, but once it happened, I went “oh, of course, that makes sense, that fits the lore that was established hours ago”.

3 Likes
14

What I really like about that AI generated definition is that it’s about how “players” think if a puzzle is logical or not.

At the other side of the spectrum, I was streaming a good point and click the other day, and sometimes I find doing seemingly logical stuff somehow was already the solution, and I was like, was this a puzzle?

Because nowadays “moon logic” is being defined by the smartsmest and the loudest streamers (who obviously do not need to read instructions, and is more interested in generating “content”), I’d say that the definition can be useful, maybe.

15

I tried to find out what this dates back to and found out it’s a pre-internet term from at least the 1970s.

However, it’s definitely adventure games that popularized it. Does anyone who was around in the 80s-90s know when it got a second wind? Was there a game that everyone started using it in relation to?

My suspicion is that tvtropes gave it a third wind in the mid-2000s, which is why its so popular now, though I don’t see it in the earliest archived tvtropes pages.

16

To be fair, this is my favorite puzzle in Adventure, but it only works because of what that game is. It makes no sense mimetically, but yet it’s completely fair, because most players do eventually try that—at least, the 1970s sort of player who would play Adventure hunched over a line printer at 2 AM, type YES out of pure exasperation, and laugh themselves silly when it actually works.

The Mellon puzzle in The Fellowship of the Ring feels similar. The puzzle isn’t finding the password; the puzzle is realizing that you were never being asked for a password.

8 Likes
17

I actually think the LLM definition in the first post gets it quite wrong and confuses several different aspects of bad puzzle design. To me, moon logic is when the elements of a puzzle are clearly part of a puzzle, clearly evident to the player, yet related/used in a way that requires very lateral or unconventional thinking. Often but definitely not always involving wordplay.

I would say the Oddly Angled Room is as close to a definitive example as you’ll find, but the dragon is also a fine demonstration.

18

I think the dragon example doesn’t fit moon logic. It’s a relatively reasonable leap of logic since you are literally prompted by the game. Moon logic is all about unreasonable leaps. I like the puzzle because it teaches you not to have assumptions about the player character and the world, both of which seem a bit magical. Within the context of the game it’s a fair puzzle.

1 Like
19

See, I’d say it was a reasonable leap of logic in one of the Phoenix games, where I’m expecting that sort of trickery—where the game might say “you’re at the bank of the river” and I have to realize I can deposit money there, because the game told me there was a bank (just in a deliberately misleading way). To me, it doesn’t feel like Adventure is that kind of game: it’s describing things in plain, straightforward language and not omitting any important details. The dragon puzzle is the one place where you have to realize the text means something other than the obvious, natural interpretation.

Which is a nice demonstration that what’s reasonable vs unreasonable can vary by player and by game! In most games I (personally) would not consider it reasonable to have to second-guess the game text. But in the Phoenix games, there’s the expectation that the game can try to trick you, but can’t outright lie (even by omission). In the Douglas Adams games, there’s the expectation that the game will lie and you should treat it as an opponent instead of a collaborator. It all depends.

2 Likes