Check it yourself.

How to read the labels

A site publishing chain data is not automatically a source of it. Most are repeating something they read somewhere else, and the further a number travels the more chances it has to go wrong. So each entry says where its data actually comes from.

Chain

Runs a node. The number comes off the chain itself.

One hop

Reads someone else's node or API and presents it.

Primary

The document, the code, or the person's own words.

Commentary

Interpretation. Useful, but check it against the rows above.

Live chain status

Note that a normal block explorer cannot help you here. Explorers running Bitcoin Core rules follow the most-work chain and will not show you the RDTS tip at all. You need something reading a node that runs the BLAKE2b code (Bitcoin Knots 29.4.1, or the 29.4.2 release candidates); a node with only consensusrules=rdts stopped at 961,639 on August 30. And an explorer showing RDTS-looking heights is not automatically the RDTS chain either: mempool.bitcoinpurity.org follows Bitcoin Purity, a separate chain that split off after block 961,636, and shows that chain's tip. There is a note on the front page about it.

• 110explorer.comChain Self-describes as a read-only explorer over a local Bitcoin Knots RDTS node, with the mainnet tip pulled from public APIs for comparison only. That is the provenance statement you want to see. It was still completing initial block download when we first checked, and on September 1 the domain redirected to a dead page. Kept for the record; do not rely on it until it answers again.
• mempool.guideChain A mempool.space instance that follows this chain block for block; every block and address link on this site points at it. Its pools chart names miners by matching coinbase tags against a hand-maintained list, so "Unknown" means unlisted, not anonymous. We have not confirmed who runs the public instance (Léo's Umbrel store ships an explorer app under its icon); cross-check its heights against your own node, as we do.
• bip110.runChain Timeline, consensus, library, coalition, census and equipment: the fullest record of the signaling period and the split, and still the place to find the whole coalition. Its dashboard did not follow the proof-of-work change: on September 1 it still showed the RDTS tip at 961,638 and a countdown at zero. Read it as history.
• bip110.org/monitorChain Miner signaling status and historical signaling rates, on the site attached to the BIP's own author. Signaling ended with the split; useful as the record of it.
• BIP-110 Fork MonitorOne hop Side-by-side Core versus Knots comparison, reading from mempool.space and bip110.run. Inherits bip110.run's stopped tip; check the date on anything it shows.
• BIP-110 CountdownOne hop Counted down to September 1. The date passed with the chain already two days old; kept only so the link history stays checkable.

Primary sources

If a claim on this site is wrong, one of these is where you prove it.

• BIP-110, the text itselfPrimary Reduced Data Temporary Softfork (RDTS), authored by Dathon Ohm, Layer: Consensus (soft fork), status Closed. Reading the header alone settles most of what gets argued about. There is no mention of proof-of-work or BLAKE2b anywhere in it.
• bip110.orgPrimary The proposal's own site, including a guide for switching chains. Distinct from bip110.run, and easy to confuse.
• bitcoin-blake2b.orgPrimary The chain's own information site: release notes, node-operator and miner pages, FAQ. It signs nothing and names nobody, which kept it off this list for a week; the Knots release notes now point to it for more information (bitcoinknots.org/learn/2026-blake2b, which redirects to it), which settles where it comes from. btc-blake2b.org and btc-blake2b.com are the same site.
• Knots PR #359, the BLAKE2b hardforkPrimary The actual implementation, by Luke Dashjr, opened August 15 as a draft and marked ready for review on August 19. Reshapes the block header, adds a 128-bit ASIC nonce space, and states that it "aims to be compatible with PoW hardware produced by several (6 IIRC) vendors." The pull request was closed on September 4 without merging, 48 comments, and stands as the review record; the code shipped in the 29.4.1 release tags, rc1 on August 21 through rc5 on September 1. Read the PR for the reasoning and the tag diffs for what actually runs. If you read one thing about the proof-of-work change, read this instead of any article about it.
• Luke Dashjr on Sia ASIC compatibilityPrimary August 13: "Support for BLAKE2b also allows BLAKE2b-Sia miners, since the latter is a strict subset of the former. Disclaimer: Have not tested yet." The clearest statement anywhere on whether existing hardware applies. Note the disclaimer is his, not ours; Connor's Lab ran the first regtest check, then testnet4, and mainnet settled it on August 30: stock-firmware Sia-class boxes mine accepted blocks, ours included.
• Luke Dashjr on the September 1 targetPrimary August 11, replying to a mining pool asking whether to start mining BLAKE2b right away: "Target is Sep 1. Need to write and properly test the code first... Until it's ready, SHA2 continues on." This is the origin of every "forks on September 1" headline you have read. Note what coverage dropped: the word target, and the entire condition after it. The chain activated on August 30, two days early.
• Luke Dashjr on the new PoW constructionPrimary August 14, draft description of the hash input: prevblock, a 128-bit nonce, then the transaction hash. The hasher cannot see height, time or version, so ASICs cannot be made to expire. Labelled a draft when written; it shipped in the release candidates and is what mainnet runs.
• Luke Dashjr on how the algorithm was chosenPrimary August 15: the four criteria behind the shortlist, with a comparison table weighing roughly two dozen candidate algorithms. First on the list: "Existing fleet of ASICs available." Existing ASICs are stated as a requirement, not a compromise, which settles most "why not RandomX" arguments at the source. Note what is not among the criteria: CPU mining.
• Bitcoin Knots releasesPrimary Release notes and binaries. 29.3.knots20260508 and newer ship RDTS directly; 29.3.knots20260507 is the same release without it. Since August 30 those releases are not enough: they stop at block 961,639. Following the chain takes 29.4.1, which carries the BLAKE2b code. The 29.4.1 release candidates: rc1 to rc3 (August 21, 22 and 27) ran the testnet4 trials, rc4 (August 30) activated mainnet at 961,640, rc5 (September 1) checkpointed that block and hardcoded the headline, and the final, 29.4.1.knots20260508, was published here on September 2 at 19:02 UTC, binaries at bitcoinknots.org; its notes list every consensus change in one place. The 29.4.2 candidates, rc1 (September 19, 15:04 UTC) and rc2 (September 20, 04:30 UTC), carry PR #419, the 45-day coinbase maturity from block 973,440; the final, 29.4.2.knots20260508, was published September 21 at 06:02 UTC, binaries at bitcoinknots.org, the candidates' signed binaries under the test directory. The 29.4.1 candidates' signed binaries stay under test.bitcoinknots.org.
• Bitcoin Knots DiscordPrimary The project's own chat, linked from bitcoinknots.org. Where the Knots developers and the people running the chain talk in real time, and where questions get answered before they get written up anywhere. Primary in the sense of people's own words, but chat scrolls away; if something said there matters, screenshot it with the date, and prefer the pull request or release notes when they say the same thing.
• Bitcoin Knots TelegramPrimary The project's Telegram group, reached from bitcoinknots.org through a warning page that reads: "Scammers are active and it is very easy to imitate other users on Telegram! Never send someone bitcoins based on a Telegram conversation!" Luke has named it, with the Discord, nostr and X, as a place the activation headline would be posted, and it was, on August 30. Same caveat as the Discord: chat scrolls away, so date and screenshot anything that matters.
• Start9 on navigating RDTS activationCommentary Practical operator-facing writeup from a node vendor.

People worth reading directly

Their own words, not somebody's summary of them. Listing someone is not agreement with everything they say, and it is certainly not their endorsement of this site. Ordered by distance from the chain, not by follower count: the people who wrote the spec and the code, then the people mining and running the pool, then commentary.

• @LukeDashjrPrimaryWrote the 2017 proof-of-work change code the current contingency is rebased from. Removed as BIP editor on August 10 over conflict-of-interest concerns tied to BIP-110, which he did not author.
• Luke Dashjr on nostrPrimaryThe same author, off X. Linked from his own site, dashjr.org, and the key was checked against that link. He has named nostr as one of the places the activation headline would be posted, and it was, on August 30; if X is down or slow, this is the fallback.
• @dathon_ohmPrimaryWrote BIP-110 itself. The specification is his, and so is bip110.org. When this site cites what the soft fork actually says, his words are the ones that settle it.
• @cguida6PrimaryChris Guida. Rebased the proof-of-work code onto a recent Knots codebase. The clearest public statements on what is and is not scheduled come from here.
• @leo_hafGitHubPrimaryLéo. Built and maintains the path most Umbrel node runners are actually using: his community store, which carries the BLAKE2b Bitcoin Knots app (at 29.4.2 final since September 21, 08:31 UTC), the DATUM gateway app and a mempool explorer app, patched for 29.4.2's difficulty field since September 20. The node this site reads the chain from runs his build. His updates to the official Umbrel Knots app have been closed unmerged since PR #6034 on September 1; the fifth, #6108 for 29.4.2, opened September 21 at 08:18 UTC, and the official store's app is still 29.4 without BLAKE2b. When we asked about the merge timeline in the Knots Discord, he answered directly and plainly. For "how do I actually run this," his repositories are where the working code lives.
• @Roughnecks110PrimaryMined the first BIP-110 signaling block at 961,632, and the blocks that extended the RDTS chain afterward.
• @Singekino_MinerPrimaryRuns Knots and mines to Ocean Pool through his own node via Datum, and posts practical walkthroughs, including how to rent hashpower and point it at a node you control. Publicly opposed to the BLAKE2b hardfork direction since August 15, and listed anyway: the criterion here is node sovereignty, not agreement with us.
• @ConnorsLabYTYouTubePrimaryConnor's Lab. Hands-on with the actual hardware: Goldshell setup guides, a Knots-plus-DATUM walkthrough on a junk PC with a pruned node, and code contributions to the BLAKE2b DATUM gateway branch. Compiled the proof-of-work change into a regtest node and confirmed stock-firmware Goldshells mine valid blocks, the first third-party test of Luke's untested compatibility claim we know of, then repeated it in public on August 21: five stock-firmware boxes mining accepted BLAKE2b blocks in the first testnet4 trial of the release-candidate code, and published the testnet setup guide two days later. When mainnet activated on August 30, his StartOS mainnet setup guide followed within hours, and the Umbrel one on September 1 (recorded on rc4; rc5 ignores its headline step). On September 19 he published a Paperclip Pool mining guide for Umbrel and StartOS. On August 14 he said he would try not to pick sides, "an opportunity for both to prosper," and has not said otherwise since; every guide, patch and rig of his since then has been on this chain. Listed either way: the criterion is building and testing in the open, not agreement with us.
• Paul Lambpaulscode.comPrimaryPackages node software for StartOS and Umbrel through his own registry at paulscode.com. Built the first install-and-go BLAKE2b ASIC test kit: Knots and DATUM regtest apps that let anyone point a Sia-class miner at the hardfork code and, when a box fails, generate a diagnostic report that feeds fixes back upstream. Also wrote the Electrum protocol extension light wallets will need to see the 164-byte BLAKE2b header, posted to the pull request on August 24, and by August 27 had it running on both ends (a patched electrs, a patched Sparrow) against live testnet4. Not upstream in either project; since August 31 he ships them as releases of his own. Says himself he is in the stay-and-fight camp rather than the hardfork camp, and shipped the tools anyway, in his words in the spirit of cooperation between the two camps. Listed for the same reason Connor's Lab is: building and testing in the open, not agreement with us.
• @GrassFedBitcoinYouTubePrimaryBitcoin Mechanic, OCEAN's Head of Communications, currently on sabbatical from the pool by his own account; the pool's own team page links this account. Listed as "Softfork Mechanic" in bip110.run's coalition snapshot; the profile now reads "Hardfork Mechanic," and the flip no longer needs inferring from a display name. He said it on camera on August 17: he is done with the SHA-256 chain and going to work on the proof-of-work change with Luke Dashjr. His August 21 follow-up adds hands-on weight: a Guix build of the first release candidate at Luke's request, and a miner he plugged in himself submitting valid BLAKE2b shares through the new DATUM and Knots builds.
• @hodlonautCommentaryCitizen journalist. Behind the CAPTURE investigation into Bitcoin Core governance, and won the defamation case Craig Wright brought against him.
• @mattkratterYouTubeCommentaryBitcoin University. Takes no sponsorships. A model for how to do this without becoming an influencer.

This list is one-sided and says so. It collects people arguing that nodes and users define the rules rather than miners. Plenty of thoughtful people take the other view and you will have no trouble finding them; they are not here because this is not a neutral directory. Note that the people listed do not all agree with each other either, and that is deliberate. It is also short on purpose: we started with the handful we actually read and would rather grow it slowly than pad it. The public coalition list at bip110.run is where to find the whole crowd. If someone belongs here, tell us.

Models we are learning from

Not sources to check claims against; blueprints for how to operate. Two projects, at opposite ends of the stack, making the same move: publish the design, let anyone build it, stay out of the sales loop.

• SeedSignerPrimary Publishes the design and a parts list instead of selling hardware, and keeps a list of community vendors with no official endorsement, no preloaded software, and a voluntary expectation that sellers give something back. The clearest working example of building something useful in Bitcoin without becoming a shop or a shill. Their own writing is also candid that donations alone have not made the project sustainable.
• BitaxePrimary What SeedSigner is to signing, Bitaxe is to mining: schematics, board files, and firmware all published, built and sold by community vendors rather than one company. It stands out precisely because so little else in mining is open; the machines around it mostly run closed firmware. That matters here because BLAKE2b has no Bitaxe yet. The ASICs the Knots draft aims to be compatible with are closed hardware today, so the open-miner gap is sitting there waiting for someone. Bitcoin Mechanic has said he intends to fill it, by reverse engineering the Sia-class ASICs and open-sourcing the firmware. Stated intent rather than shipped code, but the gap now has a named volunteer.

This page is a living document. If you have an addition, a correction, or a source we should be reading, hello@blake2bminer.com. Same address if you want to help build this.