Bitcoin Years — Pool Ledger data ================================ Which mining pool mined each Bitcoin block, and how concentrated block production was, from the genesis block to today. Updated daily; every block's label is fetched again weekly. Built by tools/pools/build.py in the Bitcoin Years repository. Licence Derived data in this folder: CC BY 4.0. Credit "Bitview (bitview.space)" and "Bitcoin Years (bitcoinyears.com)". Block -> pool labels: Bitview series block_pool, CC BY 4.0. Pool names, tags and addresses: mempool/mining-pools, MIT, (c) 2019 btc.com (copy and licence in tools/pools/list/). How a block gets its pool (Bitview's rule, BRK crates/bitview_plugin_pools, MIT) The first pool in mempool.space's known-pool list that matches one of the coinbase transaction's output addresses or, case-insensitively, its coinbase tag. No match: "unknown". Unknown blocks are never dropped: every share below is a share of ALL blocks in the window. Limits you should know - Before 2013 most pools left no tag, and the known-pool list has no payout addresses for them (DeepBit, early Slush Pool). Those years are mostly "unknown" and get no concentration figures (reliability "insufficient"). - Pools that proxy another pool's work under their own tag cannot be seen on the chain. Since about 2022 the figures here are therefore a lower bound on real concentration. - A block share is not a hashrate share: over a short window luck matters. top1_ci_* is a 95% Wilson interval. - "ambiguous": the coinbase text matches the tags of two or more pools, or our tag check disagrees with Bitview's label. Bitview's label is kept; the flag shows where the attribution is uncertain. Files blocks/YYYY.csv.gz height, time_utc, pool, ambiguous one row per block, by the block's UTC year periods.csv one row per difficulty period (2016 blocks; the last one may be open) periods-pools.csv period, pool, blocks, share weeks.csv one row per ISO week (UTC) weeks-pools.csv iso_week, pool, blocks, share days-pools.csv.gz day_utc, pool, blocks, share pools.csv one row per pool: blocks, first/last block, peak share in a difficulty period *.json the same data for the web page; meta.json holds the sources, versions and checks events.json markers computed from the data only (first time above 30/40/50%, first time in the top five, changes in the Nakamoto coefficient); no commentary renames.json pool renames found when the pool list changed (date = date the list changed) Window columns (periods.csv, weeks.csv) blocks, unknown, unknown_share, ambiguous, ambiguous_share reliability high: unknown <= 5%; medium: <= 20%; insufficient: more (concentration columns left empty) top1_pool, top1_share, top1_ci_low, top1_ci_high top3_low, top3_high share of the three largest pools. low: every unknown block counted as a separate miner; top5_low, top5_high high: all unknown blocks counted as one miner. The truth lies between (proxy pools aside). nc_high, nc_low Nakamoto coefficient: the fewest pools that together mined MORE THAN half of all blocks in the window. nc_high counts unknown blocks as separate miners, nc_low as one miner. Empty when no set of pools reaches half.