One day you wake up and the news is announcing that Bitcoin has "forked" into two different versions of itself, and that holders of the original coin now magically own coins on the new chain too. Or you read that Ethereum has undergone a hard fork and wonder whether your ETH is still your ETH. Or you hear Bitcoin Cash described as a fork of Bitcoin and wonder whether that means anything important. Blockchain forks are one of those concepts that sound far more complicated than they need to be. Once you understand what a blockchain is and why its rules matter, forks become straightforward. They are just what happens when those rules change, and the difference between hard and soft comes down to a single principle: whether the old rules and the new rules can coexist.
By CryptoAcademy Team | Published: 2026-04-14 | 20 min read read | Category: Educational
To understand what a fork is, you first need to understand what is being forked.
A blockchain is a chain of data blocks, each containing a record of transactions. Every computer connected to the network, called a node, stores a complete copy of this chain. When someone wants to add a new block of transactions, the nodes on the network check whether that block follows the rules. If enough nodes agree it follows the rules, the block is added. If it does not follow the rules, it is rejected.
These rules are the protocol. They define things like: how large a block can be, how transactions must be formatted, how miners or validators prove they have done the required work, and what counts as a valid transaction. Without these rules, nodes would not agree on what the legitimate state of the blockchain is, and the whole system would fall apart.
The protocol is code. Code can be changed. But because thousands of independent nodes are all running their own copies of this code, changing it is not as simple as clicking "update" the way you would with a phone app. When a developer proposes a change to the Bitcoin or Ethereum protocol, every node on the network has to decide whether to adopt the new rules. And this is where forks come from.
A fork is simply what happens when the rules of a blockchain change. The word comes from the same place the physical concept comes from: a path that diverges into two. When rules change in a way that requires all nodes to agree and update, the blockchain may split into two separate paths. When rules change in a way that keeps old and new nodes compatible, the blockchain continues as a single path with upgraded behaviour.
---
A soft fork is a change to a blockchain's protocol that is backward-compatible. This means that nodes running the old version of the software can still recognise and accept blocks produced by nodes running the new version. The new rules are stricter or more refined versions of the old rules, not contradictions of them.
Think of it like updating a language's grammar rules. If the new grammar makes certain sentence structures invalid that were previously valid, but does not introduce entirely new characters or symbols, then a reader who only knows the old grammar can still understand most of what is being written. They may not be able to write in the new way, but they can read it.
In blockchain terms, this means that after a soft fork:
Old nodes see new blocks as valid because the new blocks still follow the old rules, just a refined version of them.
New nodes enforce the new rules on top of the old ones.
The network does not split into two separate chains, because there is no fundamental incompatibility.
The soft fork is complete when enough of the network has upgraded to the new rules that they become the dominant standard. Old nodes are not ejected, but they are increasingly at a disadvantage because they are operating with an older understanding of the protocol.
Bitcoin has been upgraded through soft forks multiple times.
SegWit (Segregated Witness), 2017. SegWit was a major soft fork that changed how Bitcoin stores transaction data. It separated the signature data from the transaction data, which effectively increased the amount of transaction data that could fit in each block without actually increasing the block size. It also fixed a technical vulnerability called transaction malleability. Because the change was backward-compatible, old nodes could still validate new SegWit transactions. They simply saw them as a slightly different type of valid transaction.
Taproot, 2021. Taproot was a soft fork that improved Bitcoin's privacy and efficiency. It introduced a method of aggregating multiple digital signatures into one, making complex multi-party transactions (like Lightning Network channels) look identical to simple single-signature transactions on the blockchain. Old nodes still saw these transactions as valid. New nodes could process them more efficiently and privately.
The key characteristic of a soft fork is that it does not create a new blockchain or a new cryptocurrency. If you held Bitcoin before a soft fork, you hold the same Bitcoin after it. The upgrade simply changes how transactions are processed, not what currency exists.
---
A hard fork is a change to a blockchain's protocol that is not backward-compatible. The new rules introduce changes that old nodes cannot validate under their existing understanding of the protocol. Old nodes will see new blocks as invalid and reject them. New nodes will see old blocks as invalid and reject them.
This incompatibility is the defining feature of a hard fork. When old and new nodes can no longer agree on what constitutes a valid block, the blockchain splits. Nodes running the old software continue building their chain under the old rules. Nodes running the new software continue building their chain under the new rules. These two chains share identical history up to the moment of the fork, but from that moment forward they diverge completely.
The result is two separate blockchains, each potentially with its own cryptocurrency. Anyone who held coins on the original blockchain before the fork holds coins on both chains after it, because both chains inherited the same transaction history.
Think of it like a company splitting into two independent companies. Both companies were once the same entity and share the same history, but from the split date they operate completely independently with different management, different strategies, and different futures.
Hard forks happen for several reasons:
Community disagreements. When the people involved with a blockchain disagree about what direction the network should take, and the disagreement is fundamental enough that no compromise is possible, a hard fork can formalise the split.
Major upgrades. Sometimes the change a network needs to make is so fundamental that backward compatibility is impossible. Ethereum's transition from proof-of-work to proof-of-stake involved hard forks that changed the consensus mechanism at its core.
Emergency response to exploits. In rare cases, a blockchain community may use a hard fork to effectively reverse or quarantine the effects of a major hack or exploit by creating a new version of the chain where the exploit did not occur.
---
The most instructive way to understand hard forks is through the events that produced them, because the real-world cases reveal the human dimensions of what sounds like a purely technical process.
Bitcoin has a design parameter called block size, which limits how many transactions can be included in each new block. When Satoshi Nakamoto set the block size limit at 1 megabyte (later effectively expanded by SegWit to around 2 to 4 megabytes in practice), it was a design choice that prioritised decentralisation: smaller blocks mean cheaper hardware to run a full node, which means more people can participate in network validation.
As Bitcoin grew in popularity, the limited block size became a bottleneck. During periods of high demand, the network filled up, transactions slowed, and fees rose. A debate emerged between two camps.
One group argued that the block size limit needed to be increased directly. More transactions per block meant cheaper fees and faster confirmations. They believed Bitcoin's future as a payments system depended on this.
The opposing group argued that increasing the block size would concentrate mining and validation power in the hands of entities with expensive hardware, undermining decentralisation. They believed Bitcoin's value came from decentralised validation, which smaller blocks protected.
This debate simmered for years without resolution. In August 2017, a group of developers and miners who favoured larger blocks executed a hard fork, creating Bitcoin Cash with an 8 megabyte block size (later expanded to 32 megabytes). The original Bitcoin chain continued with its smaller blocks and the SegWit soft fork as an alternative scaling solution.
Anyone who held Bitcoin before August 1, 2017 automatically received an equal amount of Bitcoin Cash on the new chain. Both cryptocurrencies still exist today. Bitcoin remains the dominant chain by an enormous margin. Bitcoin Cash has its own community and continues operating. Many other Bitcoin forks followed, including Bitcoin Gold and Bitcoin SV, each representing different visions of what Bitcoin should be. Over 100 Bitcoin forks have been created since Bitcoin's launch, though most are now effectively defunct due to insufficient community support.
The Bitcoin Cash story illustrates the fundamental truth about hard forks: they are not primarily technical events. They are political events. The code is relatively straightforward. The hard part is deciding who gets to decide what the code should say.
The Ethereum hard fork of 2016 is arguably the most philosophically significant fork in crypto history, because it forced a choice between two incompatible values.
In 2016, the DAO, a decentralised autonomous organisation running on Ethereum's smart contract platform, had raised approximately $150 million worth of Ether, making it one of the largest crowdfunding events in history at that point. An attacker found a vulnerability in the DAO's smart contract code and exploited it to drain approximately $60 million worth of ETH (roughly one third of the DAO's funds).
The Ethereum community faced a choice. The original Ethereum premise, inherited from Bitcoin, was that the blockchain is immutable. The code is law. What the smart contract allows to happen is what happens, regardless of intent or outcome. Under this principle, the attacker had not broken any rules. They had exploited a vulnerability that the code permitted. The loss was legitimate.
But $60 million of real value had been taken from real people who had invested in good faith. The Ethereum community voted to execute a hard fork that effectively rewound the blockchain to a point before the exploit, redirecting the stolen funds back to their owners.
This was an extraordinary intervention. It contradicted the principle of immutability that blockchain proponents had long held as the technology's core value. Essentially, the majority of the Ethereum community decided that protecting users from an exploit mattered more than preserving the fiction that blockchains are perfectly immutable.
Not everyone agreed. Those who believed that the immutability of the blockchain should be inviolable rejected the fork and continued building on the original chain, which became known as Ethereum Classic (ETC). The new chain, which had reversed the exploit, became the dominant Ethereum (ETH) we know today.
The Ethereum/Ethereum Classic split raised questions that still resonate: if a blockchain community can vote to reverse transactions when the outcome is bad enough, what exactly is the guarantee that blockchain offers? The two chains represent two different answers to that question. ETH represents the view that community governance can override the code when the stakes are high enough. ETC represents the view that code is law, full stop, regardless of consequences.
> Real-world example:
> "Held ETH through the DAO fork period and genuinely struggled with the decision of which chain to support. Philosophically agreed that blockchains should be immutable. Practically understood that the attack had extracted real value from real people through a code bug they had not intended to create. Ended up supporting the fork because a technology that cannot address catastrophic exploits will not gain the institutional trust needed to grow. Kept a small