We've written a lot lately about what makes a question good — the anatomy, the pipeline, the difficulty band. All of that machinery exists for one reason: the questions in this post.
These are real specimens from our early boards (plus a couple of composites, labeled as such). Every one of them made it past a younger, more trusting version of the pipeline and onto a real table. We keep them the way a woodshop keeps a jar of stripped screws: partly as education, mostly as humility.
Exhibit A: the leak
Which Terry Pratchett novel, the 26th in his Discworld series, was titled "Thief of Time"?
Answer: Thief of Time.
This one is real, and it is our masterpiece. The question does not hint at its answer. It does not gesture toward its answer. It states its answer, in quotation marks, and then asks you what the answer is.
The best part — the part we genuinely could not have scripted — is that nobody got it. The table read the question, concluded that no game would simply hand over the answer, decided it must be a trick, and went looking for a cleverer response than the one sitting in plain sight. The question was so bad it wrapped back around to being difficult. We do not award points for irony, but if we did.
This specimen is why leakage is the first thing the review pass checks. A writer with its head full of facts will occasionally forget which side of the question the answer belongs on. The reviewer's whole job is to be the person at the table saying "you just... you just said it."
Exhibit B: the insult
What is the organ, located inside the skull, that is essential to cognition?
Answer: the brain.
A real 100-pointer from an anatomy category. On paper it's technically a question. In practice it's a diagnosis — get it wrong and the question itself explains why.
Here's the thing we learned from this one: the failure isn't that it's easy. The 100 is supposed to be easy — it's the handshake. The failure is that there's no path to feeling clever. A good easy question lets you arrive three ways: know it, half-know it, or reason it out. This one offers a fourth way only: possess the answer. Which everyone does. That's not a handshake, it's a form to fill in.
Exhibit C: the lecture
What was the exact displacement, in tons, of the RMS Titanic?
Answer: 52,310 tons.
(A composite — we've written a dozen of these; none survived long enough to screenshot.) This is the opposite failure: a question nobody at the table could reach by any road. No reasoning path, no half-memory to lean on, no category to land in. It isn't hard, it's unavailable — a lecture with a pause in it, graded.
The tell for this one is the answer format. When the answer is a category ("a comet"), half-knowledge has somewhere to land. When it's a five-digit serial number, the question is only playable by someone who has read one specific table on one specific page. We now check for exactly that.
Why the 100s are the funniest
Notice that our two favorite real specimens are both from the bottom of the board. That's not a coincidence, and it took us a while to understand why.
Hard questions fail in boring ways — too obscure, wrong fact, sloppy phrasing. But easy questions fail hilariously, because "too easy" is genuinely the hardest thing in trivia to calibrate. The ceiling of a topic has structure: an expert knows what an expert knows, and the 500 can be measured against it. The floor is a guess about what everyone knows — and a writer who knows everything about a topic is the worst-placed writer on earth to make that guess. Aim one notch too low and the question stops being a question and becomes a wellness check.
So when a 100 goes wrong, it goes wrong in the direction of comedy. We've made our peace with this. The review pass now asks, of every easy question, whether a player can feel clever answering it — not merely conscious.
The jar of stripped screws
Every exhibit above became a rule. Leakage checks exist because of Exhibit A. The feels-clever test exists because of Exhibit B. The answer-format check exists because of the Exhibit C family. The pipeline didn't get better by being clever in advance; it got better by being wrong in public and writing it down.
Which is also an invitation: if a board ever deals you a future exhibit, the feedback form is right there in the menu. We read every report. The best ones end up as rules — and the very best ones, apparently, end up in blog posts.
Type six topics. Statistically, at least 35 of your 36 questions will not contain their own answer.