Krillion Rarity Tiers

'Too Clever' is the second-lowest rarity tier, worth 15 points. The game gives it to answers that feel clever on the surface but are actually overthought picks that thousands of players arrive at together — a collision trap, not a rarity win. It sits between Plankton (10 pts, the most obvious answers) and Schooler (30 pts, solid middling picks).

In Krillion, an answer's rarity decides both your score and how deep you dive. The core rule is simple: common answers float while rare answers sink. The moment you submit, the algorithm checks how often your word was submitted by players worldwide — its share of all valid global answers — and sorts it into one of six rarity tiers. A word almost nobody types earns 100 points; the first word everyone thinks of earns barely 10.

Every valid answer lands in exactly one of the six official tiers below. The examples come from the official demonstration prompt "Name a landlocked country". Depth converts at 1 point = 10 metres, so each answer sinks as deep as its points.

Two real prompts show the full tier spread in action — and why the "clever" middle of the pack is where scores go to die.

• "Name a dinosaur" — One in a Krillion (100 pts): Leaellynasaura, Megalosaurus. Too Clever trap (15 pts): Archaeopteryx and Compsognathus, pub-quiz staples thousands of players also remember. Plankton (10 pts): Tyrannosaurus rex, Triceratops, Stegosaurus.
• "Name a cocktail" — One in a Krillion (100 pts): Remember the Maine, a 1930 Savoy deep cut. Too Clever trap (15 pts): Negroni Sbagliato and Boulevardier, drinks the internet spread far wider than bartenders realize. Plankton (10 pts): Margarita, Mojito, Martini, Old Fashioned.

The 15-point band is where overthinkers crash. Words that feel obscure but are famous from pub quizzes and social media — Liechtenstein instead of an obvious landlocked country, Archaeopteryx instead of an obvious dinosaur — occur to thousands of players at once, so they all collide in the same low band. To actually score, aim for the top of the table: highly specific, academic, or historically obscure words that you can also spell perfectly. Spelled correctly, a Deep Cut earns 85 points and a true One in a Krillion earns 100.

💡 If you have heard it in a bar quiz, so have thousands of other players — it is a 15-point word.

A word has no fixed rarity in Krillion — its tier is always relative to the other valid answers for that prompt on that day. "Vienna" is a Plankton answer to "name a European capital" and simultaneously a much deeper answer to "name a city with a famous music scene", because the pools of players — and their first instincts — differ between the two prompts. This is why the same word can sit in different tiers across our archive, and why tier tables like the one above are illustrative rather than fixed prices. The practical consequence: you cannot memorize a list of "rare words". You train the habit of climbing, and you judge rarity fresh against each prompt's likely crowd.

• The game compares your answer against the valid global answer pool for that prompt that day.
• Player-share bands (<1%, 1-5%, 5-15%, 15-35%, >35%) are what actually decide the tier.
• A word nobody submitted cannot be rare — if it is not on the accepted list at all, it is simply a zero. Rarity starts where validity ends.

Every answer reveal is a data point about the crowd, and consecutive reveals sketch the whole distribution. Open with a Plankton and you have located the crowd's first instinct; land a Too Clever and you have found the second cluster — the pub-quiz layer. A Schooler means you are past both clusters but still in populated water; Rare and above means you reached water the majority never swims in. Players who treat each reveal this way calibrate quickly: after a week of dives you can predict your tier before the game shows it, and that prediction accuracy is the real skill the depth meter is measuring.

💡 Predict your tier out loud before submitting. Seven correct predictions in a row means you have internalized the crowd model — the points follow automatically.