I Asked 47 Questions and Still Failed. Here's What I Learned About Solving Lateral Thinking Puzzles.

Last Christmas my niece beat me at a lateral thinking puzzle in four questions.

The puzzle was Hanging. A man is found hanged in an empty room, puddle of water under him. I spent fifteen minutes asking about rope quality, ceiling height, suicide notes, whether the door was locked. She asked: "Was there ice?" Done. The man stood on a block of ice, it melted, that's your puddle.

She's six. I've been playing these puzzles for three years.

That night I sat down and actually thought about what separates people who solve these things in five questions from people who grind through forty-seven and still miss it. Not the theory. The actual moves.

Here's what I came up with, in the order I wish someone had told me.

---

The first thing I had to unlearn

Most people (me included, for two years) treat these puzzles like a search problem. You have a thousand possible explanations, so you fire off questions to eliminate them one by one. "Was it murder?" "Was it suicide?" "Was anyone else in the room?"

This is backwards. It's also why you get to question thirty and feel like you're going in circles.

The people who solve puzzles fast don't eliminate. They locate. They ask questions that split the possibility space in half, not questions that knock off one option at a time.

My niece didn't ask "was there a chair, was there a stool, was there a ladder, was there ice." She jumped straight to the version that explains the puddle. The puddle was the anomaly. Everything else was decoration.

That was the first lesson: find the one detail that doesn't fit, and interrogate it.

Every puzzle has one. A man shoots himself after tasting turtle soup. The shooting isn't the weird part. People do awful things. The weird part is the soup. Why does the soup matter? Start there.

In Forest Clearing, fifteen dead men in a forest. The "forest" is decoration. The "clearing" is decoration. The number fifteen is the anomaly. Why fifteen, specifically, not three or thirty?

Once I started asking "what's the one detail here that makes no sense," my question count dropped by half.

---

The five questions I now ask first, almost every time

I'm going to give you the actual list. Not because it's magic, but because I spent two years figuring out that these five questions resolve maybe 70% of the "where do I even start" feeling.

1. "Is this story realistic, or does it have supernatural / sci-fi elements?"

This one question splits the possibility space in half. If it's realistic, you can trust physics, biology, and normal human behavior. If it's not, all bets are off. You need to know which game you're playing before you play it.

I wasted twenty questions on Midnight Train once because I assumed it was realistic. It wasn't. Would have saved me a lot of pain to ask up front.

2. "Does anyone die, or is someone in serious danger?"

This sounds morbid but it's practical. If yes, the puzzle is almost certainly about the cause of death or the motive. If no, it's probably about a misunderstanding, a wordplay, or a weird situation. Two completely different games.

3. "Is the timeline in order?"

A lot of the hardest puzzles are hard because the story isn't told in the order it happened. The "I" in the story might be describing events out of sequence. Once you suspect this, the whole thing opens up.

4. "Are all the important characters actually mentioned in the setup?"

The single most common trick in these puzzles is the invisible character. Someone who's crucial to the story but never named. The turtle soup story, the hanging man, the best friend. Almost every classic has one.

5. "Is there a word or phrase that might not mean what I think it means?"

Puns, metaphors, double meanings. "Turtle soup" might not be turtle soup. "Killed" might not mean killed. This is the whole trick in Pentakill, where the narrator says they've "killed" five people and feels a bit guilty about it. You have to ask what "killed" actually means here.

That's it. Five questions. Most puzzles crack open after you've asked these, because you've established the rules of the world, the stakes, the structure, the cast, and the language.

Everything after that is just drilling into specifics.

---

The thing nobody tells you about "good" questions

Here's a mistake I made for way too long. I thought a good question was one that the host answered "yes" to.

Wrong. A good question is one where either answer tells you something.

"Was it murder?" is a fine question. If yes, you know it's murder. If no, you rule it out. Either way you've learned something.

"Was he wearing a red shirt?" is a terrible question, unless you have a specific reason to suspect the shirt matters. If the answer is no, you've learned almost nothing.

The test I use now is simple: before I ask a question, I ask myself, "if the answer is no, do I know more than I did before?" If the answer is no, I don't ask it.

This one filter has probably saved me thousands of wasted questions.

---

When you're completely stuck, do this

Every experienced player has been here. You're twenty questions in, you've established the basics, and you have absolutely no idea what happened. The story makes no sense. You're starting to suspect the puzzle is broken.

It's not broken. You're just inside a frame and you can't see it.

The move is to step outside and question the frame itself. Ask things like:

• "Is the narrator reliable?"
• "Is the 'I' in this story human?"
• "Is the 'I' the same person throughout?"
• "Am I assuming something about the setting that isn't actually stated?"

I was stuck on Elevator for an embarrassingly long time because I kept assuming the narrator was an adult. The puzzle only makes sense when you realize they're not. That was a frame I had built myself without noticing.

These "frame-breaking" questions feel weird to ask, because you're questioning the puzzle itself rather than the story inside it. But they crack open puzzles that nothing else will.

---

A real example, start to finish

Let me walk you through Fatal Phone Call, a difficulty-2 puzzle that's a perfect illustration of all of this. A wealthy man calls his wife. Then his wife dies. Why?

Spoiler warning. Seriously, go play it first if you want to. It's short and satisfying. Come back after.

Okay.

First, the anomaly. A man calls his wife. Then his wife dies. The call is the anomaly. The death might be a coincidence, but the puzzle is telling you it's not. So the question is: how can a phone call cause a death?

Run through the five questions. Realistic? Yes. Does someone die? Yes, the wife. Timeline in order? Yes. Hidden characters? Probably — someone or something on the other end of that call might matter. Wordplay? "Call" could mean phone call or could mean something else, but it seems to be a phone call here.

Now the frame. What am I assuming? I'm assuming the wife heard the call, or was near the phone, or that the call was to her in the normal sense.

Break the frame. What if the call wasn't really "to" her in the way I'm imagining? What if the wife wasn't the one who picked up?

Turns out the wife was a sleepwalker. The husband called home, the phone rang, the wife (asleep) got up to answer it, walked out a window. The call didn't tell her anything. The call woke her, badly.

That's it. Three questions once you stop assuming the phone call was a conversation.

The lesson isn't "sleepwalking is the answer." The lesson is that I had a frame — "a phone call is a conversation" — and the puzzle lived outside that frame.

---

What changed for me

I still get humbled by these puzzles regularly. My niece is going to destroy me again at some point, probably soon.

But the way I fail is different now. I used to fail by grinding — forty questions, going in circles, learning nothing. Now I fail fast and specifically. Five framework questions, find the anomaly, test the frame. If I'm still stuck after that, I'm stuck on something interesting, not stuck on noise.

If you're just starting out, I'd say pick three easy puzzles and run them with the five questions as a drill. Don't try to be clever. Just practice asking them in order, and notice how much of the story resolves before you even get to specifics.

A few I'd recommend as training puzzles:

• Fatal Phone Call — difficulty 2, realistic, teaches frame-breaking about assumptions
• Meal of Death — difficulty 2, a fairy-tale remake, teaches you to ask about wordplay and cultural references
• Best Friend — difficulty 2, family drama, teaches you to hunt for hidden characters
• Forest Clearing — difficulty 3, classic black story, teaches you to find the numerical anomaly

Once those feel comfortable, go try Turtle Soup itself. It's difficulty 4 for a reason, but it's the canonical puzzle for a reason too. You'll appreciate it more after you've built the muscle.

And when a six-year-old beats you at one of these, don't feel bad. They're not smarter than you. They just haven't built the frames yet.

That, honestly, might be the biggest lesson of all.