Receipts before rankings

The desk is four tabs, and every earlier address redirects

The Data Desk's nine tabs are now four: Network at /data/network (was /data/blockspace), Mining at /data/mining (was /data/mining-economics), Pools at /data/pools (was /data/mining-pool-coordination) and Holdings at /data/holdings (was /data/bitcoin-custody), with the treasury ledger inside it at /data/holdings/treasuries (was /data/bitcoin-treasuries). This change log and the method moved to /data/method, linked at the foot of every desk page, and the Control & Exit Atlas and How Bitcoin works are guides at their old addresses. Every earlier address, and each page's share card, answers with a permanent redirect, so links in published pieces keep working, and the pool archive files keep theirs under /data/mining-pool-coordination/. The custody dataset now has one name, Bitcoin held through disclosed claims, or the claims floor for short, in place of the custody floor and Bitcoin held as a claim. No figure changed.

Fee estimates now print below 1 sat/vB; economy read 1, not 0.20

The desk read mempool.space's whole-number recommendation, which pads every rate up to a whole sat/vB, so estimates under 1 printed as 1 and the fee ladder drew equal bars. It now reads the precise estimator and prints two places under 10 sat/vB. Read together at 23:07 UTC on September 30, the whole-number endpoint gave 1 for every rate; the precise estimator gave next block 1.02, half an hour 0.67, an hour 0.35, economy 0.20 and minimum 0.10 sat/vB, so economy 1 → 0.20. The ticker's next-block fee, the Settlement Bench and the mining desk's fee readings moved with it, and the pressure reading now uses the precise next-block rate.

IBIT is counted from its 10-Q; the floor is 8.8895 percent at block 969,248

The iShares Bitcoin Trust row now carries the 734,261 BTC its quarterly report on Form 10-Q states as of June 30, with the filing on EDGAR as its receipt. It had carried 782,623.42 BTC as of September 16 from the sponsor's daily holdings file, whose link no longer serves the figure the row counted. The ETF category falls by 48,362.42 BTC, and the floor moves from 1,834,388 to 1,786,026 BTC, from 9.1302 to 8.8895 percent of the 20,091,403.125 BTC issued at block 969,248. The height and the other rows are unchanged, and the summary was recomputed through the refresh job's own arithmetic. The method now says plainly what the floor sums: each holder's latest disclosure, dated June 30 to September 27, against issued supply at the pinned block, so a holder that sold since its disclosure could hold less. Methodology 2.2.0.

Hashprice is now printed in sats per TH/s per day, at difficulty

The mining desk's hashprice is now the reward per block in sats × 86,400 × 10¹² ÷ (difficulty × 2³²): the sats one TH/s expects in a day at the current difficulty, taking the tip height's subsidy plus the average fees of the last 144 blocks as the reward. It needs no hashrate estimate and no price, and the formula is printed beside the figure. The three dollar models it replaces multiplied by a fetched price, and two of them divided by an estimated hashrate. For block 968,255, a reward of 313,369,555 sats at difficulty 132,757,073,449,487.52 gives 47.48. Mining method sg-mining.v3.

The Data Desk no longer uses a fetched price

No Data Desk figure is multiplied by a market price any more, and the desk's loaders no longer read a price endpoint. The mining desk's BTC price and dollars-to-sats tiles are gone, and the Miner's Bench now prints expected sats a day, kWh a day and a break-even power price in sats per kWh, against which a reader can set their own tariff; nothing is converted. Bench links shared before today keep their machine and network inputs and drop the dollar ones. The ticker's 1 BTC and 1 sat remain the site's only price, read through a loader of their own.

One-week sample re-read; the leaders' share rose to 61.72 percent

The top three labels hold 632 of 1,024 blocks in the one-week window, 61.72 percent, against 60.04 percent on September 18 and the highest reading the desk has recorded. The three built one block more than on September 18 in a week of 27 fewer blocks; within them AntPool rose to 223 and F2Pool to 159 while Foundry USA fell to 250. The one-month window is 59.80 percent and the three-month 59.90 percent; fifteen labels are named in the week, and six blocks carry no label. The previous snapshot and the endpoint receipts are preserved beside it.

Pool snapshots are built by script and dated by the source's clock

The pool block-share snapshot is now built from the three endpoint responses and checked against the page's own validator before anything is written; a failed or partial read writes nothing. Its observation time is the source's HTTP Date header rather than the moment the receipt was assembled, and the date in its method text is generated from that observation. The September 18 file's method text named September 12. Methodology 1.1.0; block counts and share arithmetic are unchanged.

Strategy advanced to 847,666 BTC as of September 27

Strategy's September 28 8-K reports 847,666 BTC as of September 27, with 1,665 bought that week; its September 21 8-K reported 950 bought and 846,000 held as of September 20. The row had stood at 845,050 as of August 30, and the selected total moves from 1,041,371 to 1,043,987 BTC. The filings state whole bitcoin and do not reconcile to the unit: the two purchases sum to 2,615 BTC while the filed aggregates differ by 2,616, and the ledger records the difference between the aggregates.

Strategy's filing moved the floor to 9.1302 percent at block 969,248

The corporate category rises by the same 2,616 BTC to 1,043,987, so the disclosed floor is 1,834,388 BTC, 9.1302 percent of the 20,091,403.125 BTC issued at block 969,248, against 9.1196 percent at block 967,564 on September 18. The observation height moved to a block mined on September 30, after Strategy's September 27 position, so that no counted holding is dated after the block its share is taken against. The ETF and state rows are unchanged, and the summary was recomputed through the loader's own arithmetic.

The custody floor is cut, never rounded up

The floor's share of issued bitcoin is now cut to four places rather than rounded, so it can never print above its evidence; the September 18 reading of 9.1196 percent was 9.11959 rounded up. The unclassified remainder is 100 minus the floor, and the homepage chart cuts each named part the same way. The pinned block is now required to follow the newest holding the floor counts. Methodology 2.1.0; no holding changed.

One-week sample re-read; the leaders' share rose to 60.04 percent

The top three labels hold 631 of 1,051 blocks in the one-week window, 60.04 percent, against 57.55 percent on September 12 and the highest reading the desk has recorded. Foundry USA built 275, AntPool 202 and F2Pool 154. The one-month window is 58.50 percent and the three-month 59.24 percent, so the move sits in the recent sample; sixteen labels are named in the week, and 13 blocks carry no label. The previous snapshot and the endpoint receipts are preserved beside it.

Corrected 2026-09-30 · This entry first read "seventeen labels are named in the week". The week lists seventeen rows, but one of them is the Unknown row, 13 blocks that carry no label; sixteen labels are named. The snapshot was always right.

ETF and state holdings advanced; the floor is 9.1196 percent at block 967,564

The iShares fund file moved to 782,623.42 BTC as of September 16, about 3,894 higher than the September 2 reading, and the El Salvador wallet endpoint to 7,778.37 BTC. Strategy's 8-Ks of September 8 and 14 report 845,050 BTC as of September 7 and September 13, and its September 1 8-K states no holding, so the corporate row is unchanged. The observation height moved to block 967,564 and the summary was recomputed through the loader's own arithmetic.

Corrected 2026-09-30 · This entry first said Strategy's three September 8-Ks each report 845,050 BTC, and then that the September 8 and 14 filings each report it through September 13. Only those two filings state that figure, the September 8 filing as of September 7 and the September 14 filing as of September 13; the September 1 8-K is a dividend filing with no holdings statement. The corporate row was always right.

One-week sample re-read; the leaders' share fell to 57.55 percent

The top three labels hold 610 of 1,060 blocks in the one-week window, 57.55 percent, against 59.88 percent on September 3. The one-month window is 58.72 percent and the three-month 59.00 percent. Twenty-six labels are now named rather than folded into a remainder, and the per-block receipts are published beside the snapshot.

All three windows re-read from the same capture

The top three labels hold 59.88 percent of the one-week window, against 57.59 percent on August 29. The one-month and three-month windows barely moved, so the change sits in the recent sample rather than the trend.

Corrected 2026-09-14 · This entry first read 59.86 percent. The preserved receipt for that capture is 609 of 1,017 blocks, which is 59.88 percent. The snapshot and the page were always right; the figure was mistyped here.

ETF and state holdings advanced to their newest disclosures

The iShares fund file moved to 778,729.91 BTC as of September 2, about 7,089 higher than the August 25 reading, and the El Salvador wallet endpoint to 7,763.37 BTC. The observation height moved to block 965,362.

Cango advanced from a first-quarter to a second-quarter filing

The company disclosed 1,056 BTC at June 30 against 1,026 at March 31, with 656 BTC mined in the quarter. Its two September filings concern an NYSE listing matter and carry no bitcoin figures.

Strategy and Strive observations advanced to current filings

The ledger now separates bitcoin acquired from the equity issued to fund it, and retains the share-count receipt beside the holding.

Observation-level provenance added

Each approved metric can now carry an effective date, observation date, source locator, authority tier, methodology version, and revision chain.

One-week, one-month and three-month windows synchronized

Pool labels remain a measure of block-building coordination, not miner ownership or control of hashrate.