Eight design questions

Eight choices, eight questions.

A design is a set of answers to questions somebody asked first. These eight are the ones that shape Bitcoin most. Each layer below states the choice, the worry that prompted it, what it costs, and the question it leaves standing.

The layers are stacked like a core sample, light at the top and dark at the bottom. They are in an order that builds: the chain gives a shared record, proof of work decides who extends it, the supply rule says how new coins appear, and the pace sets the rhythm. The last four are about who may take part, how, and on what terms.

01Why does Bitcoin bundle payments into a chain of blocks?
02Why does Bitcoin use proof of work?
03Why is the supply capped at 21 million?
04Why a block about every ten minutes?
05Why are payments tied to keys instead of names?
06Why is the software open source?
07Why does Bitcoin favour fixed rules over flexible ones?
08Why do transactions carry fees?

Why does Bitcoin bundle payments into a chain of blocks?

Bitcoin does not record payments one at a time. It gathers them into blocks and links each block to the one before it by its hash, so the whole record reads as a single line from the first block to the newest. That line is what lets strangers agree on one order of events, and notice at once if the past is edited.

The choice: Payments are grouped into blocks. Each block header contains the hash of the previous header, so changing any old block changes every hash after it. The payments inside a block are summarised by a Merkle tree, which lets a short proof show that one of them is included.
The worry it answers: How can strangers with no referee agree on a single order of payments, and notice immediately if someone edits the past?
What it costs: Every full node stores and checks the entire history. New payments must wait to be placed in a block, and the newest blocks can still be replaced if a competing branch gains more work.
The question that stays: How much storage and waiting is a fair price for a record nobody owns?

Descend this shaft Test it against a worry

Why does Bitcoin use proof of work?

Somebody has to be allowed to add the next block, and in an open network anyone can show up. Proof of work settles it with a puzzle: find a block header whose SHA-256 hash falls below a target. There is no known shortcut to finding one, and a single hash is enough to check it.

The choice: A block counts only if its header hashes to a value below the target. Producers change a number in the header and try again, over and over. Verifying the result takes one hash. The target is retuned so the pace stays steady as machines come and go.
The worry it answers: In an open network anyone can claim any number of identities, so counting identities cannot decide who gets to extend the record.
What it costs: Producing blocks consumes real hardware and energy. That expense is deliberate, because it is what makes rewriting history expensive, but it is a continuing cost that someone has to pay.
The question that stays: Is buying security with physical resources a better deal than buying it with a permission list?

Descend this shaft Test it against a worry

Why is the supply capped at 21 million?

The number of coins that can ever exist is not a promise; it is arithmetic. The block subsidy began at 50 coins and halves every 210,000 blocks, so the running total climbs toward 21 million and stops just short of it. Every node checks the schedule, so a block that over-issues is simply refused.

The choice: A subsidy of 50 coins per block at the start, halved every 210,000 blocks. Summed over all blocks it approaches 21 million. Transaction fees are separate, and they are not new coins.
The worry it answers: Who can create money, and can that power ever be used quietly, or in a way that holders cannot see coming?
What it costs: The schedule cannot respond to events. And because the subsidy shrinks toward nothing, fees must eventually carry more of the cost of security, which nobody can yet say will be enough.
The question that stays: Is a rule that cannot bend worth more than a policy that can adapt?

Descend this shaft Test it against a worry

Why a block about every ten minutes?

The network adjusts its difficulty every 2016 blocks, aiming for one block about every ten minutes. The target is an average. Because block discovery is a game of chance, individual gaps can be much shorter or much longer, and a payment’s first confirmation arrives on the same random clock.

The choice: Every 2016 blocks the software compares how long the last batch took with two weeks, and moves the target to bring the average back toward ten minutes per block.
The worry it answers: How quickly can the network agree without constantly splitting into competing branches that waste work?
What it costs: A first confirmation takes ten minutes on average and sometimes far longer. Anyone who needs more certainty waits for more blocks.
The question that stays: Which payments can afford to wait, and which cannot?

Descend this shaft Test it against a worry

Why are payments tied to keys instead of names?

Coins are locked to public keys, and spending them takes a signature, not an identity document. There is no registry of names. The ledger itself, though, is open to everyone, so what Bitcoin offers is pseudonymity: names are absent, activity is visible.

The choice: A coin is locked to a public key; moving it requires a signature from the matching private key. No account holder’s name appears anywhere in the rules, and the whole ledger is public.
The worry it answers: How can anyone check a spending right without an identity authority, and without a private ledger that outsiders must simply trust?
What it costs: Pseudonymity is not anonymity. Address reuse and shared inputs can link payments together, and once an address is tied to a person, its history comes with it.
The question that stays: Can a ledger be open to all and private to its users at once?

Descend this shaft Test it against a worry

Why is the software open source?

The software that implements the rules is published for anyone to read, run, copy and propose changes to. For a system whose pitch is that you need not trust a middleman, that is close to a requirement: rules you cannot read are rules you must take on faith.

The choice: The software that enforces the rules is published openly. Anyone can read it, run it, modify it, and propose changes. Which version a person runs is their own decision.
The worry it answers: How can a system that promises you need not trust a middleman ask you to trust rules you are not allowed to see?
What it costs: Reading consensus code takes skill, so most people rely on the review of others. Open projects are slow to change, and arguments can be public and bitter.
The question that stays: Who is accountable when nobody is in charge?

Descend this shaft Test it against a worry

Why does Bitcoin favour fixed rules over flexible ones?

Bitcoin’s monetary and validation rules are meant to change rarely, and only by broad adoption. Each node enforces them independently, so there is no referee to petition and no committee to persuade. That makes the system predictable, and slow to repair.

The choice: Validation and monetary rules are checked by every node on its own. Changes need broad voluntary adoption; there is no vote and no administrator who can push one through.
The worry it answers: If rules can be changed by a few, they can be changed against everyone else, and holders would have to hope those few stay well-behaved.
What it costs: Real problems may wait years for a fix that nearly everyone accepts. Disagreement can split the network into groups following different rules.
The question that stays: Where does flexibility belong: in the base rules, or in the layers built on top?

Descend this shaft Test it against a worry

Why do transactions carry fees?

Block space is limited, so someone has to decide whose payment goes in first. Fees are the answer: senders bid, block producers choose. Fees are also what the design expects to carry more of the cost of security as the block subsidy halves toward nothing.

The choice: Senders attach a fee to each transaction; block producers fill limited space with the payments that offer the most for the room they take.
The worry it answers: Without a way to ration block space, anyone could fill it with junk at no cost, and honest payments could not signal urgency.
What it costs: Fees vary with demand, so a payment can become expensive when space is scarce. And the long-run balance between fees and security spending is an assumption, not a proven fact.
The question that stays: What evidence would show whether fees can pay for enough security?

Descend this shaft Test it against a worry