Home/Findings/943 Days of Silence
Finding · free in full · coverage heights 0–960,000
943 Days of Silence
Satoshi wrote the famous sentence. Nobody wrote another for 943 days.
In block 0, the coinbase carries a printable run of 70 characters — the Times headline.
Then, for heights 1 through 139,689 (139,689 blocks, 943.6 days), the longest
readable run in any coinbase was 7 characters. Every block that even reached 7 carried the literal string
Eligius. The silence broke at height 139,690 on 2011-08-05T08:42:59Z.
Measured gapless over heights 0–960,000 (960,001 blocks), frozen at block 960,000
(2026-07-28T16:33:32Z), from our own full-archive node — no API, no third-party dataset.
Registry: F-2026-07-31-007 (silence) · F-2026-07-31-008 (BIP34 refutation), both CONFIRMED T1_CONTROLLED.
Don't take our word for it
Download verify_silence.py and run it against your own Bitcoin Core node.
Standard library only. No pip installs. No access to our machine. Exit code 0 only if your node agrees
with the headline claim.
Download verify_silence.py Method and controls Mandatory limits
python3 verify_silence.py # deterministic sample
python3 verify_silence.py --full # every block from 0 to 139,690No curl-pipe-bash. Serve and run the file as-is.
The silence
139,689 blocks. Longest readable run: 7.
Heights 1–139,689. Genesis at height 0 · 2009-01-03T18:15:05Z · longest printable run 70.
In the window: longest run anywhere = 7; blocks reaching 7 = 250 (all contain literal Eligius);
longest run excluding that tag = 4. There are no 5s and no 6s in the run histogram.
Silence is not emptiness. Those coinbases are full of bytes — binary extranonce and difficulty data. The claim is that none of it was readable text (printable runs as defined below).
Run-length histogram inside the silence window
Counts of blocks by longest printable run. Note the gap: no 5s, no 6s. Series: 0→33,575 · 1→88,678 · 2→15,039 · 3→2,112 · 4→35 · 7→250.
The break
Height 139,690 — a Latin prayer
Break height 139,690 · 2011-08-05T08:42:59Z · longest run 55:
Eligius/Benedictus Deus. Benedictum Nomen Sanctum eius.
In 2011, 134 blocks had a run ≥ 20; of those, 132 (98.5%) contained literal
Eligius. That string dominated readable coinbases for 141.2 days. The first non-Eligius block after genesis with run ≥ 20 is height
158,898 · 2011-12-24T14:20:21Z · run 20 · check out offthefed.
Genesis → first non-Eligius sentence: 1,084.8 days (2.97 years).
Quotable extra — block 142,596, 2011-08-26T01:25:33Z, run 82:
2011-08-25 18:54:55 <cjdelisle> ran out of prayers?! That explains the price drop.
Eligius is a string in the bytes, not an identification. A coinbase tag is self-reported and unauthenticated. We name no person and no company.
Finding · equal prominence · F-2026-07-31-008
The story everyone tells about BIP34 is wrong
A common claim: coinbases became unique because BIP34 made the block height mandatory in the coinbase. We tested that story two ways. It fails both.
Our own draft once implied the BIP34 story. Controls killed it. That retraction is part of the product — same standard as the Nonce Archaeology CAST retraction: wrong claims die in public, at full size.
| Month | Blocks | % distinct | % v2 |
|---|---|---|---|
| 2011-01 | 5,161 | 6.14% | 0.00% |
| 2011-06 | 6,256 | 8.12% | 0.00% |
| 2011-10 | 3,749 | 30.41% | 0.00% |
| 2011-12 | 4,585 | 53.06% | 0.00% |
| 2012-04 | 4,210 | 67.60% | 0.00% |
| 2012-06 | 4,535 | 70.87% | 0.00% |
| 2012-07 | 4,773 | 74.67% | 0.06% |
| 2012-09 | 4,695 | 78.00% | 12.27% |
| 2013-02 | 4,658 | 88.94% | 55.28% |
| 2013-04 | 4,967 | 91.99% | 100.00% |
Distinctness rose from 6.14% to 70.87% while version-2 signalling stayed at 0.00%. The uniqueness rise ran before v2.
Height-push removal test — if BIP34 height encoding were the uniqueness engine, stripping the height push should collapse distinctness. It barely moves modern years:
| Year | % distinct | Height push removed | Change |
|---|---|---|---|
| 2012 | 67.32% | 53.41% | −13.91 pts |
| 2013 | 91.95% | 85.06% | −6.89 pts |
| 2020 | 99.98% | 99.95% | −0.03 pts |
| 2025 | 99.80% | 99.44% | −0.36 pts |
We refute BIP34-as-cause. We do not name a replacement mechanism. Do not read a pet theory into the gap.
Year series · publish whole or not at all
Readable coinbase text, year by year
Coverage ends at frozen height 960,000. 2026 is a partial year. % with text is an upper bound on prose, never a measure of it — modern coinbases are dominated by pool tags. Never headline that column as “people writing more.” Distinct messages saturate mechanically after 2013 — not a cultural signal.
| Year | Blocks | Distinct | % distinct | Mean run | Max run | Carries text | % text |
|---|---|---|---|---|---|---|---|
| 2009 | 32,490 | 161 | 0.50% | 0.4 | 70 | 1 | 0.00% |
| 2010 | 67,920 | 3,625 | 5.34% | 1.2 | 3 | 0 | 0.00% |
| 2011 | 59,627 | 8,457 | 14.18% | 1.5 | 82 | 134 | 0.22% |
| 2012 | 54,526 | 36,705 | 67.32% | 10.3 | 52 | 6,607 | 12.12% |
| 2013 | 63,433 | 58,328 | 91.95% | 13.0 | 83 | 4,163 | 6.56% |
| 2014 | 58,865 | 58,253 | 98.96% | 11.1 | 73 | 5,636 | 9.57% |
| 2015 | 54,321 | 53,822 | 99.08% | 15.5 | 77 | 17,366 | 31.97% |
| 2016 | 54,851 | 53,984 | 98.42% | 15.6 | 79 | 18,038 | 32.89% |
| 2017 | 55,928 | 55,255 | 98.80% | 18.1 | 72 | 23,403 | 41.84% |
| 2018 | 54,498 | 54,462 | 99.93% | 14.2 | 52 | 12,880 | 23.63% |
| 2019 | 54,232 | 54,159 | 99.87% | 14.7 | 42 | 7,778 | 14.34% |
| 2020 | 53,222 | 53,210 | 99.98% | 15.9 | 76 | 8,998 | 16.91% |
| 2021 | 52,686 | 52,600 | 99.84% | 16.5 | 85 | 15,405 | 29.24% |
| 2022 | 53,188 | 52,992 | 99.63% | 17.9 | 82 | 20,462 | 38.47% |
| 2023 | 53,999 | 53,738 | 99.52% | 19.9 | 79 | 26,403 | 48.90% |
| 2024 | 53,473 | 53,229 | 99.54% | 20.8 | 56 | 27,679 | 51.76% |
| 2025 | 53,082 | 52,974 | 99.80% | 21.1 | 76 | 27,715 | 52.21% |
| 2026* | 29,660 | 29,633 | 99.91% | 20.3 | 48 | 14,147 | 47.70% |
*2026 partial year ending at frozen height 960,000.
Max printable run by year (published series)
Values: 70, 3, 82, 52, 83, 73, 77, 79, 72, 52, 42, 76, 85, 82, 79, 56, 76, 48 (2009–2026*).
Negative result · real billing
The 2016 halving was marked. The 2024 halving was not.
| Height | Text | Run |
|---|---|---|
| 420,000 | Chandler Guo loves YangYang Jin. | 32 |
| 420,001 | /BTCC/ Welcome to 12.5 BTC blocks! BTCC & Bitcoin Forever! | 58 |
| 840,000 | /ViaBTC/Mined by buzz120/ | 26 |
Controls · the product
Six controls. All passed.
| Control | Result |
|---|---|
| C1 threshold sweep | Genesis alone in 2009 at every bar 3–70; 2010 empty at every bar ≥ 4; break block identical for bars 8–50 |
| C2a independent re-read | 41 blocks vs Core raw hex, 0 mismatches |
| C2b exhaustive end-to-end | 139,691 blocks re-read from the node, 279,383 RPC calls, claim confirmed without our database |
| C3 coverage | 0 empty coinbases in the window; rendering length min 5 / mean 26.7 / max 44 |
| C4 instrument | 7 synthetic cases, 7/7 exact |
| C5 BIP34 timing | Refuted the draft — rise 6.14% → 70.87% at 0.00% v2 |
| C6 BIP34 mechanism | Height-push removal moves 2020 by −0.03 pts, 2025 by −0.36 pts |
Mandatory limits
Why you should believe the rest
- Coinbase
scriptSigonly.OP_RETURN, output scripts and inscriptions are not measured. “Nobody wrote anything” means “in the coinbase field.” - Silence is not emptiness. Those coinbases are full of bytes — binary extranonce and difficulty data. The claim is that none of it was readable text.
Eligiusis a string in the bytes, not an identification. A coinbase tag is self-reported and unauthenticated. Name no person and no company.% with textis an upper bound on prose, never a measure of it. Modern coinbases are dominated by pool tags. The rise from 17% (2019) to 52% (2025) is mostly branding getting longer. Never headline this number as “people writing more”.distinct messagessaturates mechanically after 2013. Not a cultural signal.- We do not claim what did make coinbases unique. BIP34 is refuted; no replacement is named. Do not let a page imply one.
- Coverage is 0–960,000. No claim about later blocks.
- Do not mix with the registry metric
coinbase_text_rich(LENGTH(ascii) > 60, used by the Curiosity Engine year-step atoms). Different measure, different numbers. Its σ values are not about this series.
Method
Definitions
- printable byte = 0x20..0x7E
- printable run = a maximal consecutive stretch of printable bytes in the coinbase scriptSig, with leading/trailing whitespace removed
- longest run = the length of the longest such stretch in that block
Source: our own full-archive node. Frozen height 960,000. Registry anchors
F-2026-07-31-007 and F-2026-07-31-008, CONFIRMED T1_CONTROLLED.
Reproduce against your node: verify_silence.py.
This finding is free-only
No fourth product. Three sealed datasets already exist. Free kits on those pages: samples, manifest, OpenTimestamps, verify.py before pay.
Download verifier Sealed datasets (other findings) How seals work
Cite
Citation
Copy for papers and notes. Prefer the free finding URL. Attribute figures to PARALLAX only as published on the linked page; do not invent numbers. Carry the limits section with any citation of the headline claim.
BibTeX
@misc{parallax-943-days-of-silence-2026,
author = {PARALLAX},
title = {943 Days of Silence},
year = {2026},
howpublished = {\url{https://parallaxbtc.com/findings/943-days-of-silence.html}},
note = {Coinbase scriptSig printable-run silence from height 1 to 139,689 (943.6 days) after genesis; BIP34 uniqueness claim refuted. Coverage heights 0--960,000. Free verify_silence.py against the reader's own node. Pseudonymous research; figures re-derivable.},
url = {https://parallaxbtc.com/findings/943-days-of-silence.html}
}
RIS
TY - ELEC
AU - PARALLAX
TI - 943 Days of Silence
PY - 2026
UR - https://parallaxbtc.com/findings/943-days-of-silence.html
N1 - Coinbase scriptSig printable-run silence heights 1-139689 (943.6 days); BIP34 uniqueness refuted; coverage 0-960000; free verify_silence.py.
ER -
Plain URL: https://parallaxbtc.com/findings/943-days-of-silence.html