How does Bitcoin let many computers agree on what gets recorded?

Each computer can check a payment - but we still need to decide what gets added to the record

Suppose Emma has enough money for only one payment.

She tries to use the same money once to pay Liam and once to pay Olivia.

Each computer can check a request for itself: is it authorized, are the funds still available, and does it follow the rules?

But if both requests use the same money, only one can be added to the record.

So checking a transaction is not enough. We also need a shared way to decide which transactions are added to the record - and in what order.

Transactions enter the record inside blocks

A transaction is a request to change the record. For example: Emma sends Liam 30 USD.

Bitcoin does not add every transaction to the record one by one. Several transactions are collected together, along with information that connects them to what was already recorded.

That package is called a block.

A block is a package trying to become the next update to the record.

But if there is no manager choosing the next update, who creates it?

Miners try to create the next block

Some participants run computers that try to create the next block. They are called miners.

Each miner prepares a possible block and repeatedly tries a computation connected to it.

Each attempt produces a new result, and nobody can know in advance which attempt will meet Bitcoin's requirement.

A miner that finds a qualifying result can publish its block to the network.

This computational search is called Proof of Work.

More computing power means more attempts, which gives a greater chance of finding a qualifying result.

Finding the result gives the miner a chance to propose a block. It does not give the miner authority to change the rules.

Nodes still check every block

When a miner publishes a block, nodes on the network check it for themselves.

They check that its transactions follow the rules, that already-spent money is not being spent again, and that the Proof of Work is valid.

If something is invalid, they reject the block.

So there are two different roles:

Miners try to create and propose the next block.

Nodes check whether the block can be accepted.

Even a miner with a great deal of computing power cannot make an invalid transaction valid.

Each new block connects to what was already recorded

A new block does not stand alone.

It contains a reference to the block before it. The next block refers to that one, and so on.

This creates a connected sequence of blocks.

That sequence is called a blockchain.

In simple terms, each block says:

“I continue from here.”

Sometimes two valid continuations appear almost at the same time

Information takes time to move from computer to computer across the network.

One miner can find and publish a block.

Before another miner hears about it, that miner can also find a valid block that continues from the same earlier block.

For a short time, there are two possible continuations of the same chain.

Both blocks can be valid.

This temporary split is called a fork.

So which branch should the nodes continue from?

Nodes continue from the heavier branch

Every block requires Proof of Work.

When new blocks continue to build on one branch, more computational work accumulates behind that branch.

Among valid branches, Bitcoin defines the preferred chain as the one with more accumulated Proof of Work.

You can think of it as the heavier branch.

“Heavier” does not necessarily mean simply having more blocks. It means more work has accumulated behind it.

A block on the other branch was not necessarily fake or invalid. It simply did not remain part of the main chain that the network continued from.

An orange Bitcoin chain briefly splits into two valid branches. One branch accumulates more Proof of Work and continues in orange. The other block stays visible, turns gray and moves aside. The preferred branch is determined by accumulated work, not simply the number of blocks.

As more work accumulates, changing the past becomes harder

Suppose Emma's payment is already inside a block on the chain the network is following.

More work builds after it, and then more again.

To replace that history, it is not enough to edit an old record. An alternative history would need enough Proof of Work to overtake the history the network is already following.

So as more work accumulates after a payment, confidence in that payment generally grows.

This is also where the idea of confirmations comes from.

There is no magic number that makes a payment 100% impossible to reverse. Confidence increases gradually.

So what is Bitcoin doing?

Nodes check transactions and blocks for themselves.

Miners compete through Proof of Work for the chance to propose the next block.

Blocks connect to one another to form a blockchain.

And if two valid continuations appear temporarily, nodes continue from the branch with more accumulated work.

That is how many computers can eventually converge on the same record - without one organization managing it.

What if we want the network to do more than record who paid whom?

So far, the rules have checked payments and helped decide the order in which they are recorded.

But what if we could also decide in advance what should happen to money when a condition is met?

For example:

People send money to a project. If 10,000 USD is raised by a certain date, the money goes to the project. If the goal is not reached, the money is returned to the people who sent it.

What if computers on the network could carry out that rule themselves?