About

Decision paralysis costs time. When two options carry comparable weight, cognitive load increases without yielding better outcomes. This generator uses the browser's crypto.getRandomValues() API to produce a single unbiased random bit - the digital equivalent of a fair coin with P(A) = P(B) = 0.5. Unlike Math.random(), which relies on pseudo-random number generators (PRNGs) seeded by predictable system states, cryptographic randomness draws from OS-level entropy pools. The distinction matters in aggregate: over 10,000 trials, PRNG bias can skew results by up to 0.3‰. This tool logs your decision history locally so you can audit distribution fairness yourself.

Limitation: randomness does not account for option quality. If one choice carries asymmetric risk - financial, safety, or otherwise - rational analysis outperforms any random method. Use this tool for genuinely equivalent alternatives where speed of decision matters more than optimality. Pro tip: if you feel disappointed by the result, that reaction itself reveals your true preference.

Formulas

The selection algorithm generates one cryptographically random byte and maps it to a binary outcome:

R ← crypto.getRandomValues(Uint8Array[1])

choice =

{

A if R[0] < 128B if R[0] ≥ 128

The byte R[0] is uniformly distributed over [0, 255]. Splitting at 128 yields exactly 128 values per outcome, so:

P(A) = 128256 = 0.5

Where R = raw random byte from the OS entropy pool, A and B = user-provided choices, P(A) = probability of selecting choice A. The split at 128 is exact because 256 = 2⁸ is evenly divisible by 2, eliminating modulo bias entirely.

Reference Data

Method	Entropy Source	Bias Risk	Speed	Use Case
Fair Coin Toss	Physical mechanics	~51% same-side bias (Diaconis et al.)	~2s	Casual decisions
Math.random()	PRNG (xorshift128+)	Low but deterministic seed	<1ms	Games, non-critical
crypto.getRandomValues()	OS entropy pool	Negligible (<2⁻¹²⁸)	<1ms	Security, fair selection
Dice Roll (d6, pick odd/even)	Physical mechanics	Manufacturing variance ±0.5%	~3s	Board games
Atmospheric Noise (random.org)	Radio atmospheric noise	Unmeasurably low	200−500ms (network)	Lotteries, research
Quantum RNG (ANU)	Quantum vacuum fluctuations	Theoretically zero	100ms (API)	Cryptography, auditing
Lava Lamp Wall (Cloudflare)	Chaotic fluid dynamics	Negligible	Continuous feed	TLS key generation
Mental Coin Flip	Human cognition	High (cognitive biases)	Instant	Not recommended
Rock-Paper-Scissors	Human psychology	High (pattern exploitation)	~5s	Social, informal
Drawing Straws	Physical arrangement	Moderate (position bias)	~10s	Group decisions
Shuffled Deck Card Draw	Physical shuffle quality	7 riffle shuffles for uniform	~30s	Card games
Binary Hash of Timestamp	System clock LSB	Periodic, predictable	<1ms	Weak fallback only

Frequently Asked Questions

The tool reads one byte from crypto.getRandomValues(), which taps the operating system's entropy pool (e.g., /dev/urandom on Linux, CryptGenRandom on Windows). The byte range [0, 255] splits exactly in half at 128: values 0 - 127 select option A, 128 - 255 select option B. Since 128 ÷ 256 = 0.5 exactly, there is zero modulo bias.

Math.random() uses a pseudo-random number generator (typically xorshift128+ in V8). Its output is deterministic given the seed, and the seed is derived from system time at engine startup. For binary decisions this bias is negligible in isolation, but over thousands of trials, patterns may emerge. The Web Crypto API draws from OS-level entropy (hardware interrupts, mouse movements, disk timing), making the output cryptographically unpredictable.

Yes. The history panel logs every decision with a timestamp. Over a large sample (e.g., 100+ choices), you can count outcomes for each option. A fair generator should converge to 50% per option. If you observe a deviation greater than ±5% over 200 trials, something is anomalous - but statistically, fluctuations of ±3% are expected.

Partially. Research by Ap Dijksterhuis (2006) on unconscious thought theory suggests that for complex decisions with many attributes, allowing the subconscious to process options yields better satisfaction. The coin-flip disappointment reaction acts as a somatic marker - a gut signal that reveals preference before conscious reasoning catches up. It is not a replacement for analysis when stakes are high, but it reliably surfaces hidden preferences.

Random selection is appropriate only when E(A) ≈ E(B), where E is the expected value of each outcome. If one choice involves irreversible consequences (financial loss, health risk), use expected utility theory or a decision matrix instead. This tool is designed for decisions where both options are genuinely acceptable.

No. All data is stored exclusively in your browser's localStorage. Nothing is transmitted over the network. Clearing your browser data erases the history permanently. The tool operates entirely offline after initial page load.