Chapter 4 · Why Bitcoin Was Invented

A Real-World Problem

You take a photo and send it to a friend. Afterward, the photo remains on your phone.

This is a basic property of the digital world: copying costs almost nothing, and sending means copying. Music, documents, videos, and code all work this way.

Why, then, are money transfers different?

When you send your friend 100 yuan, your balance falls by 100. The money is moved rather than copied. But the numbers on the screen are also just data. Why can they not be copied?

Chapter 2 gave the answer: an institution subtracts and adds on a single ledger. The institution guarantees movement; the data itself has no special property.

Now remove the institution.

Suppose a file says, “This is a digital coin, serial number 000731.” You send it to A. The file remains on your computer, so you also send it to B.

A has a copy, B has a copy, and you have a copy. All three are identical. Nobody can say which is the real one.

The question is:

How can the same money be prevented from being spent twice in a digital world without a central institution?

Computer science discussed this problem for more than twenty years. Chapter 3 located the difficult layer: ordering, not forgery. This chapter examines the specific answer offered in 2008.

Thought Experiment

Move chapter 1's village online and try successive versions.

Version one: shells become files.

The village turns each shell into a file, such as shell-000731.txt. Whoever holds the file owns the shell.

It fails within three minutes. Copy and paste turns one into two. This version cannot even prevent forgery.

Version two: add chapter 3's secret mark.

Instead of files, people send notes. Everyone has a private seal only they can use. A transfer note says “Number 000731 belongs to B” and bears the owner's seal. Everyone can verify the owner stamped it.

Forgery is solved: without the seal, you cannot produce a valid note.

But you can still write “000731 belongs to B” for B and “000731 belongs to C” for C, stamping both with your genuine seal.

Each note is valid in isolation. B and C each verify their own note and believe they received the shell. This is chapter 3's conflicting announcements in standard digital form.

Version three: use one public ledger and accept whichever is recorded first.

The direction is right, but “first” returns the original problem. The network has no objective order: B sees their note earlier, while C sees theirs earlier. Who announces the order? Whoever does becomes Wang from chapter 2.

Version four: make announcing the order require real expenditure.

The rules become:

1. Everyone collects received notes in a basket.
2. Anyone may copy a batch onto a page and append it to the previous page.
3. For others to accept that page, its author must complete a difficult task: a puzzle solved only by repeated guesses, whose correct answer everyone can immediately verify.
4. The village accepts the longest sequence of pages. To overturn an earlier page, an attacker must redo the work from that page onward, catch up with everyone else's accumulated work, and exceed it.

Return to version two's double spend. Both notes enter the basket, but a page author may copy only one. Including both invalidates the page because the same shell appears twice, which anyone can detect.

The conflict is resolved because only one note can enter the longest sequence, not because someone judges which note is true.

Notice the ingenuity. This does not establish which note was physically earlier—that question has no answer. Instead, it makes overturning established order require continued expenditure.

Your Decision

The village must formally choose who may write the next page. Four proposals follow.

What Happens

The proposals appear to select writers but actually do the same thing:

Put an unforgeable price on the authority to announce order.

Compare them:

The last two columns matter most. A sustainable rule needs both a price that cannot be conjured from nothing and a real financial loss for wrongdoing.

Now return to the initial question: how do we prevent money being spent twice? The answer contradicts most people's intuition.

Not by protecting the data. It can still be copied freely; anyone can copy your transaction record.

By making everyone agree on order and making that order expensive to overturn. Only one of your two notes enters the longest sequence. The other is not destroyed; it is simply never accepted.

Read that several times. It is Bitcoin's design core, and explains why the system must be slow, costly, and reward its participants.

Build a Model

A transfer takes this path before becoming effective:

The second model explains long-term stability through a negative feedback loop:

1. More puzzle solvers join → pages arrive faster
2. Periodically, the system checks whether recent pages arrived faster or slower than the target
3. Faster pages make the puzzle harder; slower pages make it easier
4. Production returns to its target rhythm

One counterintuitive consequence deserves attention:

Doubling network computation does not increase page production speed. It makes the puzzle twice as hard while maintaining the rhythm.

In other words, computation buys the probability of writing during a period and the expense of rewriting history, not speed. Beginners often misunderstand this; the AI lab tests it specifically.

The third model concerns scarcity. Physical scarcity comes from matter itself. Digital systems lack that condition and need three supports:

The third is easily overlooked but underpins the first two. If only a few supercomputers can verify the ledger, you must trust their claims about whether rules changed secretly. This is why a full node running on a home computer matters so much: it determines your position on chapter 3's trust scale.

Finally, how are assets recorded? This system does not store account balances. It records individual unspent notes:

• Receiving money adds a note addressed to you.
• Spending invalidates whole old notes and creates a new note for the recipient and a change note for yourself.
• Your balance is the sum of your unspent notes, not a number stored somewhere.

One direct benefit: checking whether money was spent only requires checking whether its note remains in the unspent set, not replaying all history. Double-spend verification becomes a lookup.

What It Is Called

Now name the mechanisms.

Bring it together:

Bitcoin is the first digitally scarce asset without an issuer, rather than merely a better payment method.

Real resource expenditure buys agreement on one history among strangers without an administrator.

Hands-on

AI Lab

Describe a miner’s work in no more than twenty lines of pseudocode.
Include candidate block assembly, what is attempted repeatedly, the success condition,
what is broadcast after success, and how other nodes verify it.

Use logic only, not specific library or language APIs.
Add an English comment to every line explaining what it does.

Assume network hashrate permanently doubles tomorrow and all other conditions remain unchanged.

Question 1: In the long run, what fraction of its previous value will average block time become?
Question 2: Will blocks arrive faster before the next difficulty adjustment?
Question 3: How does doubled hashrate affect the cost of overturning an existing block?

Show the derivation for every answer, not just the conclusion.

Real-World Case

Change One Variable

Takeaway Questions

Choose three of the Eighteen Questions for this chapter.

Chapter Check