Behind the Spin: How Digital Random Wheels Generate Provably Fair Results

Explore pseudo-random number generators (PRNG), modulo arithmetic, friction curves, and client-side entropy in digital wheels.

Why Standard Math.random() Is Not Enough for Fair Draws

Many casual web spinners rely on Math.random() paired with simple rounding (Math.floor(Math.random() * n)). While fast for visual demos, naive modulo operations across non-power-of-two ranges introduce slight modulo bias—meaning earlier slices can become marginally more probable over thousands of spins.

How Cryptographic Rejection Sampling Guarantees Exact Uniformity

Every spin on our random wheel uses the browser's hardware-backed window.crypto.getRandomValues() API (a 32-bit unsigned integer pool) combined with rejection sampling:

1. Compute a Divisible Cutoff: Calculate limit = Math.floor(0x100000000 / n) * n so the accepted range divides evenly by the number of choices n.
2. Reject Out-of-Range Samples: Draw a 32-bit random integer; if it exceeds limit, discard it and re-sample before computing rand % n.
3. Separate Selection from Animation Physics: Once the winning slice is selected mathematically, the HTML5 Canvas animates 5 to 8 full rotations with a smooth quintic ease-out curve (1 - (1 - t)5) so the wheel slows naturally onto that exact wedge without altering odds.

Why Client-Side Execution Matters

Because all draws execute inside local browser memory (or on our roulette wheel simulator), there is zero network latency and zero server-side log exposure of participant names.