A bookseller in Portland recently ran an A/B test on her store's reading-challenge app. One group of customers earned a gold star for finishing a themed puzzle — match the first line to the novel, say. The other group earned a badge reading "First-Line Detective." Same puzzles, same difficulty, same audience. The badge group came back and completed a second puzzle 31% more often. The question worth sitting with: why would a label outperform a symbol so decisively, and what does that tell us about how readers actually decide to keep playing?
The Star Is a Score. The Badge Is an Identity.
A star is a tally. It tells you that you did well, then it files itself away in a row of identical stars. A skill badge tells you what you are. Behavioral economists call this a shift from outcome framing to identity framing, and it shows up everywhere from fitness apps to language software. The badge carries a claim: you have a particular competence, and that competence has a name.
This matters because readers who see themselves as "a person who solves literary puzzles" behave differently from readers who see themselves as "a person who has accumulated 14 stars." The first group has a self-concept to protect. The second group has a spreadsheet.
Variable Rewards Explain the Return, Not the Label
B.F. Skinner's work on variable-ratio reinforcement is the standard explanation for why puzzles hold attention at all. When the payoff arrives unpredictably — sometimes the answer comes in ten seconds, sometimes in four minutes — the behavior persists longer than it would under a fixed schedule. Book puzzles are naturally variable. You don't know how hard the next one will be.
But variable reinforcement only explains why people start and continue in the moment. It doesn't explain the 31% gap between two groups getting identical variable rewards. That gap lives in the label, not the schedule.
Loss Aversion Does the Rest
Here is where Daniel Kahneman and Amos Tversky's loss aversion becomes useful. People weigh losses roughly twice as heavily as equivalent gains. A star you haven't earned yet is a potential gain. A badge you have earned is something you now possess — and a badge collection with a gap in it reads as a loss.
The Portland bookseller noticed this in her retention data: badge holders who had earned three of five "genre detective" badges returned at noticeably higher rates than star holders with the same three completions. The unfinished set did the work. The star row, by contrast, never felt incomplete, because stars are fungible. Three stars is just three stars. Three badges out of five is a story with a missing chapter.
What This Means for How You Design the Next Puzzle
If you run a bookstore, a library program, or a reading community, the practical lever is specificity. "Mystery Solver" beats "Level 3." "Poetry Sorter" beats "50 points." Name the skill, not the score.
There's a risk worth naming: badges can tip into competitive play that alienates casual readers. Research on competition and intrinsic motivation — Deci, Koestner, and Ryan's 1999 meta-analysis is the standard citation — found that controlling, pressuring rewards can undercut genuine interest, while informational feedback tends to support it. A badge that says "you have this skill" is informational. A badge that says "you beat 400 other people" is a different animal, and it tends to shrink the pool rather than deepen it.
The forward-looking move for booksellers is to treat badges as a form of reader self-description, then build the next puzzle to extend that description. If someone earns "First-Line Detective," the follow-up shouldn't be a harder first-line puzzle. It should be a puzzle that lets them become something adjacent — "Unreliable Narrator Spotter," maybe. Stars cap out. Identities compound.