Loading interactive controls…

UUID Collision Lab

A tiny space. A surprising duplicate.

Predict it. Draw it. Challenge someone else.

TOY IDENTIFIERS · LEARNING MODEL

How soon is sooner than you think?

Choose a small space, guess the first repeat, then draw.

THE DRAW TRAY

First duplicate on draw 8

8 draws · 7 distinct toy IDs

Fixed-seed example shown. Reset to make your own prediction.

Chance of a repeat by draw 8: 87.92%

Model: independent, uniform draws with replacement from 16 values.

A seeded replay illustrates the idea. It does not test real UUID collisions.

Keep the experiment in perspective.

1Predict first

The surprise is in the reveal,
not a screenshot of the answer.

2Change the space

Compare 16 and 256 values.
Watch how the scale changes.

3Keep it a model

UUID v4 has 122 random bits.
This is not a UUID stress test.

Questions, Answered.

Is this generating real UUIDs?

No. It draws small integer IDs in a toy space to explain collisions.

Why can a repeat happen before the space is full?

Each draw is independent, with replacement. It may land on any earlier value. The chance of at least one repeat grows as more earlier values can match.

Does the seed change the probability formula?

No. The seed changes the repeatable illustration. The displayed formula assumes independent uniform draws with replacement, not a measurement of the seeded sequence or real UUID failures.

What does my friend receive?

A settings-only link with version, bit count and seed. Their prediction is blank and replay starts at draw zero. No sender prediction, progress, answer, title or production UUID is sent.