Randomness As a Service

Security verification

Verification

Secure randomness, backed by public verification.

This page summarizes the standards references, conformance checks, and published statistical results behind the service.

Reference conformance15 statistical checksPublic verification

Current snapshot

Verified

Reference conformance

3/3 matched

Independent reference cases are included in the published verification set.

Published samples

4 large + 32 repeated

The verification page publishes both large-sample and repeated-run evidence.

Statistical coverage

15 statistical tests

4 large verification samples plus 32 repeated verification samples are published here.

Observed balance

50.006% / 49.994%

Distance from 50-50: 0.006%.

Visual inspection

Bitstream visualization.

A clean generator should read as texture rather than stripes, grids, or repeating bands. This view renders a fixed verification sample directly from the generator output so visible structure would stand out immediately.

Verification sample

The bitmap shows one bit per pixel. The trace below it shows how each row moves around the 50% midpoint instead of settling into a fixed pattern.

1 bit per pixel

Row balance trace

Bitmap

256 x 256

Each pixel maps directly to one output bit from a published verification sample.

Rendered sample

65,536 bits

8 KiB of generator output rendered without post-processing.

Observed split

50.081% / 49.919%

Light pixels represent one bits. Dark pixels represent zero bits.

Reading the bitmap

Visible bias usually appears as banding, checkerboards, repeated diagonals, or blocks with obvious spacing. None of those structures should dominate a healthy sample.

Visual inspection is supplementary. The formal evidence on this page remains the reference-conformance checks and the published statistical results below.

Evidence

Public verification overview.

The service is checked against independent reference cases and a broad statistical battery, with outcomes published here.

Reference conformance

3/3 matched

Independent reference cases used during verification.

Statistical coverage

15 tests

Published results from both large-sample and repeated-run verification.

Bit balance

50.006 / 49.994

Observed ones-versus-zeros split across the published verification samples.

Verification scope

Public summary

Reference conformance

3/3 matched

Independent reference cases are part of the published verification set and are included alongside the statistical battery.

Verification coverage

  • 32 repeated verification samples
  • 4 large verification samples
  • 15 statistical test families with published pass counts and p-value ranges

Statistical verification

The published verification results below show all 15 test families in one ledger. Each row includes sample scope, pass count, and observed p-value range.

Alpha 0.01

Distribution

Bit balance and run-behavior checks across repeated published samples.

4 tests

Frequency (Monobit)

Repeated verification samples

32/32 samples0.0329 .. 0.9160

Frequency (Block)

Repeated verification samples

32/32 samples0.0743 .. 0.9873

Runs

Repeated verification samples

32/32 samples0.0109 .. 0.9703

Longest Run of Ones

Repeated verification samples, block size 128

32/32 samples0.0154 .. 0.9854

Structure

Matrix, spectral, and complexity checks over large verification samples.

4 tests

Binary Matrix Rank

Large verification samples

4/4 samples0.1739 .. 0.7211

Discrete Fourier Transform

Large verification samples

4/4 samples0.0534 .. 0.9960

Linear Complexity

Large verification samples

4/4 samples0.0416 .. 0.8026

Cumulative Sums

Large verification samples

4/4 samples0.2438 .. 0.9621

Pattern

Template, entropy, and compressibility checks across large samples.

5 tests

Non-Overlapping Template Matching

Large verification samples

4/4 samples0.0034 .. 0.9973

Overlapping Template Matching

Large verification samples

3/4 samples0.0030 .. 0.5329

Maurer's Universal

Large verification samples, block length 8

4/4 samples0.0893 .. 0.9171

Serial

Large verification samples

4/4 samples0.1335 .. 0.9611

Approximate Entropy

Large verification samples

4/4 samples0.1984 .. 0.9783

State transitions

Random-walk style checks when the sequence contains enough return cycles.

2 tests

Random Excursions

Large verification samples

3/3 applicable0.0077 .. 0.9809

Random Excursions Variant

Large verification samples

3/3 applicable0.0125 .. 0.9889

Coverage

What this verification covers.

This section summarizes what has been verified, the scope of the current evidence, and the standards references behind it.

Verified properties

  • Public verification includes matched reference cases and a published 15-test statistical battery.
  • Published results cover both repeated-run sampling and large-sample verification.
  • Observed bit balance stays close to 50-50 across the published verification samples.

Limits

  • This page summarizes verification outcomes rather than disclosing internal generator construction details.
  • This repo is not FIPS 140-3 or CMVP validated.
  • Statistical checks and reference conformance are strong engineering evidence, not a formal unpredictability proof.

Source links

External references
NIST SP 800-22 Rev. 1aCMVP / FIPS 140 validation