Block Reorganizations: When Blockchains Rewrite Their Recent History

If you've ever sent Bitcoin to an exchange and wondered why you have to wait thirty minutes before you can trade it, you're bumping up against one of the blockchain's most misunderstood features: the possibility that what just happened might unhappen.

When most people think about blockchains, they imagine something permanent, carved in stone the moment a transaction is broadcast. But the truth is more nuanced and, honestly, more interesting. Blockchains have something like short-term memory that can be rewritten, and only after enough time passes does that memory solidify into something we can truly trust. This phenomenon is called a block reorganization, or a reorg, and understanding it means understanding why blockchains work the way they do.

Think of a blockchain not as a single chain everyone agrees on, but as a competition between possible chains. At any given moment, thousands of miners or validators across the world are racing to add the next block. Most of the time, one miner finds a valid block, broadcasts it, and everyone else accepts it and starts building on top of it. Clean and simple. But sometimes, two miners find valid blocks at almost the same moment. Maybe one is in Singapore and one is in Iceland. Both broadcast their blocks. Now the network has temporarily forked into two possible futures.

Here's where it gets interesting. The miners who hear about the Singapore block first will start building on top of that one. The miners who hear about the Iceland block first will build on that one. For a brief moment, there are two competing versions of reality, two possible next chapters in the blockchain's story. The network doesn't panic. It doesn't need a committee to decide which one is correct. Instead, it follows a simple rule: whichever chain grows longer first becomes the canonical chain, the official history. When one side pulls ahead, everyone on the shorter chain abandons it and switches to the longer one. The blocks that were on the abandoned chain get orphaned, and any transactions in those orphaned blocks that weren't also in the winning chain go back into the memory pool to be mined again.

This is a natural reorg, and it happens all the time, usually affecting just one block. Most users never notice because their transactions end up in the winning chain anyway. But this mechanism reveals something profound about how blockchains achieve consensus: they don't achieve perfect agreement instantly. They achieve eventual agreement through competition and majority hash power.

Now let's talk about why this matters for you as a user. Imagine you send one Bitcoin to an exchange. The transaction gets included in block number five hundred thousand. From your perspective, it's done. You see it confirmed. But the exchange doesn't credit your account yet. Why? Because they know that block five hundred thousand might get reorganized. If another chain overtakes the current one, block five hundred thousand might be replaced with a different block five hundred thousand, one that doesn't include your transaction. Your transaction would suddenly become unconfirmed again, back in limbo.

This isn't paranoia. It's math. The probability that a block will be reorganized decreases exponentially with each additional block built on top of it. One confirmation means your transaction is in the latest block. Two confirmations means there's one block on top of it. Six confirmations means there are five blocks on top of it. To reorganize a transaction with six confirmations, an attacker would need to build a longer alternative chain starting from before your transaction, which requires controlling more than fifty percent of the network's mining power for a sustained period. On Bitcoin, with six confirmations taking roughly an hour, this is extraordinarily difficult and expensive.

That's why exchanges typically wait for multiple confirmations. Bitcoin exchanges usually want six. Ethereum might require twelve or more, because Ethereum blocks come faster but are also slightly easier to reorg in the short term. Smaller cryptocurrencies with less hash power might require dozens of confirmations, because reorganizing their chains is cheaper and more feasible for an attacker.

And yes, people do attack this way. It's called a fifty-one percent attack, and we've seen it happen on smaller blockchains. Ethereum Classic, Bitcoin Gold, and Vertcoin have all suffered deep reorgs where attackers deliberately created alternative chains to double-spend coins. The attacker deposits cryptocurrency to an exchange, waits for it to be credited, trades it for something else or withdraws a different currency, then releases a pre-mined alternative chain that doesn't include the original deposit. If the attacker's chain is longer, the network accepts it, the deposit vanishes from history, and the exchange is left holding nothing while the attacker walks away with the withdrawal.

This sounds almost like science fiction, like time travel where someone changes the past. But it's not magic. It's economics. Maintaining enough hash power to rewrite history is expensive. On Bitcoin, it would cost millions of dollars per hour to rent enough mining equipment, and you'd be competing against the entire honest network. The attack only makes sense if what you're trying to steal is worth more than what you're spending to reorganize the chain. For Bitcoin, that threshold is absurdly high. For a coin ranked two hundredth by market cap, it might be just a few thousand dollars.

There's another kind of reorg that's less malicious but equally important to understand: the reorg caused by network latency or temporary splits. If a large portion of miners temporarily loses connection to the rest of the network, maybe because of an internet backbone issue or a Great Firewall hiccup, those miners might build several blocks on an outdated chain. When connectivity is restored, the two chains reconcile, and one side's blocks get orphaned. During the 2013 fork caused by a database bug, Bitcoin experienced a reorg several blocks deep, and transactions that people thought were confirmed suddenly weren't. That was a moment of genuine uncertainty, resolved by miners coordinating to return to a common chain.

Modern blockchains have implemented various mechanisms to reduce reorg risk. Ethereum, after transitioning to proof of stake, introduced finality gadgets that make blocks economically final after a certain number of validator attestations. Once a block is finalized, reorganizing it would require destroying a massive amount of staked Ether, creating a financial disincentive that goes beyond just computational power. Other chains use checkpointing, where periodic blocks are marked as irreversible by some central or federated authority, though this trades off decentralization for stability.

The key insight here is that blockchain immutability isn't instant. It's probabilistic and strengthens over time. When someone says blockchains are immutable, what they really mean is that blockchains become immutable enough, given enough confirmations, that reversing them is economically irrational. The system doesn't rely on perfection. It relies on incentives, on making honesty cheaper than dishonesty, and on the accumulation of work or stake that buries history deeper and deeper until digging it up becomes impossibly expensive.

So when you're waiting for those confirmations, you're not waiting for some bureaucratic process. You're waiting for mathematical certainty to compound. You're waiting for enough miners to stack enough work on top of your transaction that undoing it would require burning more resources than any rational attacker would spend. You're watching economic finality emerge from computational competition.

Understanding reorgs means understanding that blockchains are living systems where consensus is constantly being negotiated, not static databases where truth is written once. The chain you see right now is just the longest chain at this moment. Tomorrow's longest chain will almost certainly include all of today's blocks, but in those first few minutes, there's always a possibility, however slight, of revision.

See you Tuesday.

Block reorganizations remind us that on a blockchain, the past isn't fully written until the present has moved far enough beyond it.