https://parallaxbtc.com/ Verifiable Bitcoin research from a self-hosted full archival node. Two views, one true measurement. Free findings in full; sealed datasets optional. en Thu, 30 Jul 2026 12:00:00 GMT https://parallaxbtc.com/findings/nonce-archaeology.html https://parallaxbtc.com/findings/nonce-archaeology.html Thu, 30 Jul 2026 01:00:00 GMT The luckiest block ever mined needed 1,082 tries. In 2010 the mean winning nonce fell to 0.122 of the range midpoint, with 96.33% of winners in the bottom half, because the original client counted upward from zero and a CPU never reached the rest of the space. It returns to uniform at height 90,000 - the arrival of GPU mining, datable from a field nobody was recording. A fainter deviation is measurable today and is pool-dependent: F2Pool 24.2 sigma against MARA at -0.9 in the same months, which places the cause in equipment rather than protocol. Includes a published retraction: a claimed universal octile depletion was a rounding artefact and does not exist. https://parallaxbtc.com/findings/scaling-war.html https://parallaxbtc.com/findings/scaling-war.html Thu, 30 Jul 2026 00:00:00 GMT Bitcoin XT signalled 41 blocks in total against a 75% threshold. Bitcoin Classic peaked at 5.41%, 77% of it from two pools. SegWit sat pinned between 24 and 36% for seven months where CSV went start-to-finish in three. Taproot locked in 41 days, then its bit decayed for five years and is still set in ~2% of blocks. Read from 960,044 blocks on our own node. https://parallaxbtc.com/findings/sleeping-fleet.html https://parallaxbtc.com/findings/sleeping-fleet.html Wed, 29 Jul 2026 00:00:00 GMT Every Satoshi-era mining reward classified coin by coin. 34,185 of 196,479 outputs have never been spent; 157,766 departures dated to the block. Includes the outpoint-level trap that makes script-level analysis silently wrong.