Clusterflick

K-Nearest Neighbors. Machine learning that isn't GenAI. Place training data, flick photo tokens, classify by proximity.

2–4 Players 30 Min Ages 14+

Interactive Tutorial. Learn to Play Beta

What you'll learn

Beyond the Black Box

ML is a family - GenAI is just one branch

Distance, Not Probability

K-NN classifies by similarity, no neural net needed

Data IS the Model

Every sample stays visible - no hidden weights

AI Fluency

Know which ML tool fits which job

Play now

Create Game

Share the room code to invite players, or play solo vs a bot.

Your Name

Choose Your Role

Player

Join a Game

Have a room code? Enter it below to join.

Your Name

Choose Your Role

Player

For organizers

Event Hub. Run Multiple Tables

Create and manage multiple game rooms at once. Players join using the shareable link for each table. You are added as an observer to each room.

Enter your name above, then click Add Table.

Running a workshop?

Room Code

----

Share the code or link with other players to join

Goal: 8+ confidence for all 6 animals to win (6 rounds max).
Your Turn: 5 flicks from your corner zone - drag back like a slingshot.
📷 Identify: Token stays where it lands. Scores 5 / 3 / 2 / 1 pt by confidence band.
🎲 Add Samples: Once per round. If token lands in an animal zone, place that many sample squares in that zone, then the token is removed. After placing, you get a free Reflick of any identified token.
Sample Walls: Sample squares block tokens - flicked discs bounce off them.
Sample Bonus: +1 confidence per matching-animal square touching your token or in the same band.
Unclassified: Centre = 0 pts.
Doubled: One animal's points 2× each round (gold ⭐).
Board clears after every round.

----

Round Schedule

Your Turn

Waiting for game to start…

Confidence Scores

Clusterflick

Create Game

Join a Game

Event Hub. Run Multiple Tables

End Game?

Game Over!