On Sunday, March 23, at 4:49:35 PM, AntPool found a valid block at height 941,881. Twelve seconds later, at 3:49:47 PM, Foundry USA found another one, at the same height. Two perfectly valid blocks, with perfectly valid transactions, competing for the same spot in the chain.

Developer 0xB10C wrote: “We just had a rare two-block fork/reorg between Foundry and AntPool+ViaBTC”. The news spread. Industry headlines spoke of “chain split”, “clash between mining pools”, “warning sign”. Some evoked the specter of 51%.

Yet, what happened on Sunday is described in detail in section 5 of the Bitcoin whitepaper. Satoshi had predicted it in 2008 - and the protocol handled it exactly as designed.

Let’s take a step back.

When a miner finds a valid block, they broadcast it to the network through what’s called the gossip protocol: every node that receives the block verifies it and forwards it to the nodes it’s connected to, which do the same, and so on. The block propagates in concentric waves through the network - thousands of nodes distributed across the entire planet.

This propagation takes time. Not much, usually a few seconds are enough. In the meantime, however, on the other side of the world, another miner might have found a valid block at the same height, without knowing that someone else found one an instant earlier.

And that’s exactly what happened on Sunday. AntPool and Foundry found two valid blocks 12 seconds apart. Both legitimate, both containing valid transactions, both with correct proof-of-work. Two right answers to the same question.

At that point Bitcoin’s blockchain temporarily split: some nodes received AntPool’s block first, others received Foundry’s. Each started working on the version they had seen first. No error, the protocol was working.

I quote from section 5 of the whitepaper:

“If two nodes broadcast different versions of the next block simultaneously, some nodes may receive one or the other first. In that case, they work on the first one they received, but save the other branch in case it becomes longer. The tie will be broken when the next proof-of-work is found and one branch becomes longer; the nodes that were working on the other branch will then switch to the longer one.”

The rule is simple: every node works on the chain it received first, but keeps the other. When one of the two branches becomes longer, all nodes converge on it. The chain with the most cumulative proof-of-work wins. Always.

On Sunday, the race went one block further than usual. ViaBTC extended AntPool’s chain to block 941,882. Foundry extended its own version of the same block. Two parallel chains, two blocks deep each - the famous “two-block reorg” that made headlines.

Then, from block 941,883 to 941,886, Foundry found four consecutive blocks. Its chain became the one with the most accumulated work. The nodes therefore decided to converge toward that chain. AntPool and ViaBTC’s blocks became orphans - valid in themselves, but discarded from the main chain because the competing branch was heavier.

And the transactions contained in the orphan blocks? Back in the mempool. Reincluded in subsequent blocks. No transactions lost, no double spend, no damage.

The entire situation resolved itself in a few minutes, without human intervention.

A one-block reorg happens roughly once every two weeks. Two miners find a block almost simultaneously, the network splits for a few minutes, the next block breaks the tie. Routine - so routine that no one writes about it.

A two-block reorg is different. It requires the tie to survive an entire additional block cycle: after the first split, a second block must be found on each branch before the network converges. The probability is drastically lower. In Bitcoin’s history - over 940,000 blocks, 17 years of operation - two-block reorgs number in the order of a few dozen.

To understand why it’s so rare, just think about the numbers. A block is found on average every 10 minutes. Block propagation across the network takes a few seconds. The window in which two miners can find a valid block without knowing about each other is therefore tiny compared to the interval between blocks. And for the reorg to reach two blocks, this coincidence must repeat twice in a row.

On Sunday there was a factor that made the event slightly less improbable than usual: difficulty had just dropped by 7.76% - one of the most significant drops of the year - due to global energy stress related to the Strait of Hormuz crisis. With less hashrate on the network, blocks are found slightly less uniformly, and the probability of temporal coincidences increases by a fraction.

But even so, it remains a rare event. The point is that “rare” and “unexpected” are two different things. The Bitcoin protocol is designed to handle exactly this situation. The reorg is the conflict resolution mechanism - the way a decentralized network, without an arbiter, reaches consensus when two valid answers compete for the same spot.

When a bank database has a conflict - and it does, more often than banks admit - human intervention is required. A technical team that identifies the problem, decides which version of the data is correct, applies the fix, verifies that no damage occurred. Hours, sometimes days. Sometimes the conflict isn’t even discovered in real time.

When Bitcoin has a conflict, the network resolves it on its own in a few minutes following a rule written in the code 17 years ago. Without calling anyone, without support tickets, without a committee meeting to decide who’s right.

On Sunday, the Bitcoin protocol handled a conflict on a global scale - thousands of nodes, hundreds of miners, millions of dollars in transactions - and resolved it autonomously, transparently, and verifiably by anyone. No traditional financial system can say the same.

Leave a comment

Morning Blend by Atlas21 is here!

From Monday to Friday, at 7:00 AM sharp, the newsletter that brings you the most relevant news of the day on Bitcoin, technology, and finance. Subscribe for free here:

Personal recommendation.

I know many of you use Relai because it was a sponsor of this newsletter for a long time. Bull Bitcoin is not a sponsor. I have no commercial agreements with them and I’m not paid to write these lines.

When you ask me where to buy bitcoin, my first answer is always the peer-to-peer market: Bisq, HodlHodl, Robosats. But I know not everyone wants to use these services. If you’re looking for an immediate service that’s compatible with Bitcoin’s original values, my choice today is Bull Bitcoin. By signing up with the code “federico” you get a reduced spread of 1.75% instead of 2%, forever. You can do it here.

Non-custodial, supports Bitcoin on-chain, Lightning, and Liquid. And you can also spend your bitcoin to pay any IBAN in euros, without going through a bank.

The difference with Relai is a matter of principle: Bull Bitcoin doesn’t collaborate. KYC data is on self-hosted infrastructure, not shared with tax agencies, governments, or third parties. And they will continue not to collaborate until someone shows up with a court order in hand. It’s the only service in Europe I can say this about with certainty.