Fair Instagram Giveaway Algorithm — Verifiable Random Winner Selection

Why SHA-256 makes the draw verifiable

SHA-256 produces a 256-bit fingerprint of the entire input set — comments, timestamp, winner count, exclusions. The hash function is collision-resistant (NIST FIPS 180-4): there is no known way to construct two different inputs that yield the same hash. So the published seed_hex is a cryptographic commitment to the exact input the algorithm ran on.

If a host swaps the winner after the draw, they must also alter the comment list — which produces a different SHA-256. Followers re-running the algorithm from the published hash will detect the mismatch instantly. Fairness becomes a mathematical property, not a promise.

seed_hex = SHA256(canonical_json({
  post_url:     "https://www.instagram.com/p/<shortcode>/",
  comments:     sorted([c.lower() for c in comments]),
  timestamp:    "2026-06-11T14:23:45.000Z",
  winner_count: 1,
  exclusions:   sorted([e.lower() for e in exclusions]),
})).hex()

What "fair" actually means in algorithmic terms

A fair Instagram giveaway algorithm must satisfy four properties at the same time. Drop any one and the result is no longer auditable:

• Determinism. Identical input must always yield the identical winner. Non-determinism (e.g. Math.random()) makes the result unverifiable.
• Uniformity. Every eligible comment must have exactly 1/N probability. HMAC-DRBG + Fisher-Yates guarantees this under the random oracle model.
• Tamper-evidence.Any change to inputs must propagate to the published hash. SHA-256's avalanche property delivers this: one flipped bit changes ~50% of the output bits.
• Public auditability. Anyone, at any time, must be able to re-run the algorithm and arrive at the same winner. This requires open-source reference code + publishable inputs.

How HMAC-DRBG + Fisher-Yates produces a uniform winner

The 256-bit SHA-256 hash seeds an HMAC-DRBG instance (NIST SP 800-90A). HMAC-DRBG produces an unbounded stream of cryptographic pseudo-random bytes that is statistically indistinguishable from true randomness under standard hardness assumptions. Those bytes drive a Fisher-Yates shuffle on the sorted comment list. The first entry after applying exclusions is the winner.

drbg     = HMAC_DRBG(seed=seed_hex)         # NIST SP 800-90A
shuffled = fisher_yates(comments, drbg)     # uniform permutation
winners  = [c for c in shuffled if c not in exclusions][:N]

Why Fisher-Yates specifically? Because it is the only in-place shuffle that provably produces every permutation with exactly equal probability when fed a uniform RNG. Naïve sort-by-random-key shuffles are biased; Fisher-Yates is not.

How a follower verifies the winner is correct

Every Rafflecopter draw exposes the seed_hex on its result page. A skeptical follower has three independent ways to confirm the winner:

GET /api/verify/{seed_hex}
-> 200 { canonical_json, winner: "@ana.silveira", verified: true }

Frequently asked questions