Guess the celebrity questions: clues as constraints, uniqueness test and ordering rungs

Take a single question apart and it turns out to be a machine for shrinking a crowd. Before the first clue, the answer could be anyone the room has heard of. Each clue removes some of them. The question is finished when exactly one is left, and it is fair when the room could have followed the same path.

This page is about that machine: how clues cut, how to order them, and how to test that the set describes one person and not three.

What to take away

A clue is a constraint. Its job is to remove candidates, and a clue that removes nobody is decoration.
The set of clues must fit exactly one person the room could name. Test that against the room, not against your own memory.
Order clues from the one that cuts least to the one that cuts most, and the question becomes playable at every rung.

Clues as constraints

The children's game of twenty questions works because each yes-or-no answer roughly halves what is left. A guess-the-person question runs the same process in the other direction: the writer supplies the answers to questions the players never asked, and the players work out what remains.

Clue placement by narrowing power

Broad clue

Field removed: Very little
When to place: Early
Example: General kind of work
Risk: Wastes a rung

Narrow clue

Field removed: A great deal
When to place: Late
Example: Fits a handful
Risk: Gives game away

Seen this way, every clue has a measurable job. Before the clue, some number of people fit everything said so far. After it, fewer do. The difference is what the clue contributed.

A clue nearly everyone in the field shares, such as the general kind of work a person does, removes very little. Put it early. A clue that fits a handful removes a great deal. Put it late.

Write each clue and then ask: who is still in? If you cannot answer that roughly, you do not yet know what the clue does.

The uniqueness test

The question is defective if, after the last clue, two people still fit. This is the single most common fault in this format, and it is invisible to the writer because the writer starts from the answer and reads every clue as pointing at it.

Uniqueness test decision

After last clue, do two people still fit?

Yes

add a clue that separates them

is the one person nameable by the room?

Run the reverse test by giving the clues, without the answer, to someone who knows the field. Ask them to list everyone who fits the words, not who they think you mean.

If they name two people, the question needs a clue that separates them. If you cannot think of one, the two careers are too alike for the format, so make the question about someone else.

The test has a second half. The one person who fits must be somebody the room could name. A set of clues that uniquely describes a person nobody present has heard of is logically perfect and useless as a question.

Ordering the rungs

A descending-clue question is only playable if the clues get easier as they go. That means ordering by how much each clue narrows the field, least first.

Position	What the clue should do	Who answers here
1	Cuts the least, so a wide field stays in play	Anyone who follows the field
2	Cuts the first clear group	Players who know the era
3	Cuts about half of what is left	Players who know the career
4	Cuts all but a few names	Devotees of the person's work
5	Cuts the rest, leaving one name	Anyone who has followed the ladder

Ordering the five rungs

Firstnarrow to broad category, true of many
Secondremove most of the category
Thirdleave a short list
Fourthleave one or two
Fifthconfirm rather than narrow

Points fall as the rungs descend, which gives a table a real decision at each step: answer now for more, or wait for certainty. That decision is the pleasure of the format. It disappears if rung three gives the game away, because then nobody waits and nobody risks anything.

The difference between a question and a riddle

Some clue sets do not describe a person; they encode one. A pun on a name, a wordplay on a title, a chain of associations only the writer would make. That is a riddle, and riddles are a fine format with their own audience, but they test a different skill from knowledge and should be labeled as such.

The tell is whether a player who knows the person well could have answered from the clues. If the clues make sense only after the reveal, the question was a riddle, and the reaction in the room will be a groan rather than a nod.

Writing the set

A practical sequence for one question:

Writing one question

Choose an answer the room knows
List checkable facts about the work
Sort by how many others share each
Pick five spanning the range
Test on someone who does not know
Record the source for each clue

Choose the answer and check that the room could name them.
List everyone else who fits the general field before any clue is written.
Write the broadest clue first and note which of those candidates it removes.
Add each later clue only if it removes someone the earlier clues left.
Run the uniqueness test with a reader who knows the field.
Record the source for every clue.

Step six is the one skipped, and it is what lets you defend a clue when a table says it is wrong.

One worked example. The answer is Meryl Streep, and each clue removes candidates the earlier clues left in play.

Works in film. Removes stage-only and music-only performers.
American. Removes performers from every other country.
Born between 1930 and 1960. Removes American actors outside that window.
Has won three acting Academy Awards. Leaves a short list that includes Meryl Streep, Jack Nicholson and Frances McDormand.
Played the editor of a fashion magazine in a 2006 film. Removes everyone except Meryl Streep.

The uniqueness check: hand the five clues, without the answer, to a reader who follows film, and ask for everyone who fits. Only Meryl Streep survives, and the room can name her.

Where this connects

The general rules for keeping a question to one defensible answer are in guess the celebrity, and the uniqueness test above applies that idea to a set of clues rather than a single sentence. Clues are drawn from a body of work, and the checking problems for a long credit list are described under career quizzes.

A ladder built on screen work needs the version awareness set out in movie and TV trivia, where a remake and its original are treated as separate answers. The last rung can carry a still instead of a phrase, using the reproduction rules under photo quizzes.

Common questions

How many clues should a question have?

Five is the standard. Three is a written question with a preamble. Seven is a monologue.

What if a table answers correctly on the first clue?

Give them the points. A format that rewards nerve has to pay out when nerve wins.

Can clues include something the person said or did outside their work?

They can, and the room will feel it. Clues from work are about something the person chose to make public; the others are not, and a question that leans on them buys a point at the cost of the mood.

How do I make the first clue interesting without giving anything away?

Make it specific and broad at once: a detail that is true of many people in the field but that only a devotee would think to apply.

Should I read all five clues even after someone has answered?

Yes, briefly. The rest of the room wants to know how they would have done.