P2Pool

P2Pool is a Bitcoin mining pool with no operator. Forrest Voight announced it on the BitcoinTalk forum on 17 June 2011, and it began testing on the live network a month later. In an ordinary pool a company collects each block reward and pays the miners afterwards. In P2Pool the miners were paid by the block itself, so nobody ever held their coins.

A miner ran the P2Pool software beside their own full node, and the P2Pool nodes formed a peer-to-peer network of their own. Miners who did not want to run one could point their machines at a public node run by somebody else. Together the nodes kept a second, much smaller blockchain called the share chain. A share is a block that would be a valid Bitcoin block except that it only meets an easier difficulty target. The share chain adjusted that target so that somebody in the pool found a share about every 30 seconds, and each node kept only the latest 8,640 of them, three days' worth.

Each share carried a ready-made coinbase transaction that paid the miners who had found the recent shares before it, in proportion to their work. Now and then a share was also good enough to meet Bitcoin's own target. That share was a real Bitcoin block, and its coinbase paid everyone at once, straight to their own addresses. The miner who found it received an extra 0.5 percent of the reward, there to discourage anyone from keeping a winning block back, and the other 99.5 percent was divided by recent work.

This design removed several things a miner otherwise has to trust. No operator could run off with the pool's funds or quietly underpay, because no funds were ever held. There was no central server to knock offline. Each node built its own blocks, so no single operator chose which transactions the pool mined. And because the score counted only recent shares, pool hopping gained nothing, which is the "Hop-Proof" in the title of the announcement thread.

The same design set its limits.

• Lost shares. A chain that moves twenty times faster than Bitcoin's leaves little time for a share to reach everyone. Shares that arrived late did not count, and miners with slower connections lost more of them than the rest.
• A rising bar. The more hash power the pool had, the harder each share was to find. A small miner could go a long time without finding one, and a miner with no share among the recent ones was paid nothing from a block.
• Crowded payouts. Every miner's payout was a separate output in the coinbase, taking up block space, and the more miners there were, the smaller each payout got. To keep payouts from turning into dust, the software made a small miner's shares harder to find, so that a share it did find could be expected to earn at least 0.001 BTC when the pool found a block. The small miner's payouts stayed spendable and came less often. The code also limited a coinbase to the 4,000 largest payouts, with a comment that the limit was "unlikely to ever be hit".

The operator of a public node described the bind in the project's thread in November 2015: "If 100 PH/s was pointed to P2Pool today most (if not all) of the existing miners would see a huge spike in variance, and a large reduction in payout." (That operator now runs this site.) Pooling exists to smooth out a miner's income, and a pool can only do that in proportion to its size, so a pool that punishes its own growth cannot do the job well. The P2Poolv2 project's comparison says the same about the payouts: paying miners out of the coinbase put hard economic limits on how large the pool could grow.

The chain holds a record of all this. Every P2Pool coinbase carries an output for donations to the developer, even when almost nothing is in it, and counting the blocks that have it gives 2,545, the first on 23 August 2011. P2Pool found 1,024 blocks in 2012, 649 in 2013, 489 in 2014, 230 in 2015 and 46 in 2016. For a while the number of miners paid in each block went the other way: a typical block paid about 200 addresses in 2012 and 388 in 2014. The most any block ever paid was 553, on 19 May 2013 (block 236,962), in a coinbase transaction of 19,033 bytes. P2Pool never came near its own 4,000 limit.

The dust rule shows in the same data. In July 2014, 1,749 of the month's 15,865 payouts were between 0.001 and 0.002 BTC, about one in nine, and none was under 0.001. The line did not hold in every period (in 2017 about one payout in seven was under it), but across all 656,494 payouts fewer than one in 200 was. What the chain cannot show is the miners who earned nothing.

The last release, 17.0, came out in August 2017 and made P2Pool compatible with SegWit. The last change on the main branch of its code is dated September 2018.

ChainQuery, which is run by the same person as this site, lists more than 17,000 addresses that it attributes to P2Pool miners as of November 2026. The list is built from the coinbase transactions of the blocks P2Pool found and from the logs of the public node at p2pool.org.

A few nodes still run it. Blocks in P2Pool's format have turned up about once a year since 2020, the latest on 27 June 2026, and that one paid 52 addresses.

As of November 2026 at least two projects are trying to rebuild the idea. P2Poolv2 is a rewrite that lets a share credit other shares found at about the same time, so fewer are wasted, and pays only the largest miners in the coinbase while smaller miners sell their shares for bitcoin. Braidpool collects the whole reward in one output that a large group of its miners must sign for together. Neither has settled how a pool with no operator pays thousands of small miners out of one coinbase transaction without opening a new way to cheat them. That problem now has a name, the decentralized coinbase payout protocol, and so do the attacks on it: payout settlement griefing and the rich list attack.

See Mining rabbit hole section 8 for where mining pools stand today.