
—
Where, exactly, does your money live?
I’ve been asking friends this question for a couple of weeks, and the answers fall into two camps. Camp one says “on the blockchain, obviously.” Camp two says “in my wallet.” Both camps are wrong, and the way they’re wrong is the single most under-explained thing in crypto. Because there is no such thing as a Bitcoin balance. Not on the chain. Your wallet computes a number for you and shows it on a screen — that number is a story, not a record.
Underneath that story, Bitcoin and Ethereum keep two completely different kinds of books. Bitcoin keeps cash: a pile of discrete, unspent notes of arbitrary denomination. Ethereum keeps a bank ledger: a row per account, balances updated in place. And this one design choice — not block size, not fees, not marketing — explains most of why the two chains evolved the way they did.
This is a comparison piece, so I’ll be disciplined about it: what each model is, what each one buys you, and what each one costs you. No tribalism. Both models are correct answers to different questions.
Cash in Envelopes: How Bitcoin Records Money
That’s the opening of the Bitcoin whitepaper, and it’s the most literal sentence in the document. A bitcoin is not a number in an account. It’s an output of a previous transaction — a sealed envelope with an amount written on it and a locking script saying who may open it. Spend it, and the envelope is destroyed forever; new envelopes are created in its place.
Every Bitcoin transaction is the same machine: it consumes one or more existing envelopes (inputs) and creates new ones (outputs). Nothing is edited. Nothing is updated. Coins are consumed and re-minted in every single transaction.
So where’s your balance? Nowhere. It’s this:
"Your balance" = Σ (all unspent outputs whose locking script your keys can satisfy)
Concretely: suppose you’ve received three payments — 0.02, 0.05, and 0.01 BTC, each to a fresh address. Your wallet says “0.08 BTC.” What actually exists are three separate envelopes. When you send 0.06 BTC to a friend, your wallet picks envelopes #1 and #2 (0.07 total), destroys both, and creates two new envelopes: 0.06 locked to your friend, and ~0.0098 locked to a brand-new address of yours called the change address. Envelope #3 sits untouched.
| Piece | What it is | What happens to it |
|---|---|---|
| Input | Reference to an existing unspent output | Destroyed on spend — permanently |
| Output | A new amount + locking script | Becomes a future input, or sits unspent |
| Change | Excess from inputs, returned to sender | Goes to a fresh address, not the original one |
That change-address detail sounds like plumbing. It isn’t. It’s the privacy model. When you pay with a $100 bill for a $3 coffee and get $97 back, the cashier doesn’t learn your bank account number. Bitcoin works the same way — the model is cash, and cash doesn’t remember who held it.
The Bank Book: How Ethereum Records Money
Ethereum’s book is the one that feels familiar. The chain maintains state: a mapping of address → account, where each account has a nonce (a transaction counter, which is what prevents replay attacks), a balance, and — if it’s a smart contract — code and storage.
When you send 1 ETH, the protocol does not destroy envelopes. It finds your row in the state, subtracts 1 from your balance, finds the recipient’s row, adds 1 to theirs. Two cells, edited in place. That’s the whole transaction.
state[sender].balance −= 1; state[recipient].balance += 1; state[sender].nonce += 1
If Bitcoin is a wallet full of notes, Ethereum is a bank book with a row per account, plus a law that says every account can also be a program. That last clause is doing enormous work, and we’ll get to it.
Same Question, Nine Different Answers
Now put the two books side by side. This table is the article; everything after it is commentary.
| Question | UTXO model (Bitcoin) | Account model (Ethereum) |
|---|---|---|
| Where is state? | The UTXO set — every unspent output | One global account tree |
| “Balance” is… | Computed by wallets from the UTXO set | A stored field, read directly |
| Double-spend defense | Structural: an input is spent or it isn’t | A per-account nonce counter |
| Spending costs | Grows with number of inputs consumed | Flat per transaction |
| Privacy | Cash-like; change addresses by default | Address reuse links history |
| Validation | Transactions touch only their inputs — parallelizes | Stateful ordering per account |
| Smart contracts | Scriptless-ish; Bitcoin Script is stateless | Native; state lives in accounts |
Why the Difference Isn’t Cosmetic
Here’s where I stop describing and start arguing. The ledger model is not an implementation detail. It’s the decision that determined what each chain could become. Three consequences, in order of how much they surprised me when I first internalized them:

The Bill That Comes Due
Every model sends an invoice. Bitcoin’s arrives through the input port.
Remember: every envelope you spend adds bulk to your transaction. A native SegWit input costs roughly 68 virtual bytes. A simple 1-input-2-output transaction is about 141 vB. But a transaction spending ten small envelopes — say you collected ten small payments and now want to consolidate — is about 535 vB. At the same fee rate, that’s nearly four times the cost, for the same “one payment.” Your balance doesn’t know this. Your fees do.
And the network pays too. The UTXO set — every envelope in existence — is currently around 166 million entries, roughly 11 GB that every full node must keep on fast storage to validate anything. It peaked above 180 million during the 2023–2025 inscription era and has come down since, but the direction over the years is unmistakably up. About 3.77 billion outputs have been created since genesis. Almost all are spent. The survivors are the money supply.

For Your Reference
I’ve written a lot of words about two data structures. Here’s what I’d actually keep:
- “Balance” is a UI convention. Bitcoin computes it; Ethereum stores it. Knowing which one you’re holding changes how you reason about fees, privacy, and custody.
- Dust is a liability, not an asset. Ten tiny UTXOs spend as one expensive transaction. If you receive many small payments, consolidate them during a low-fee weekend — future you pays for present you’s tidiness.
- Neither model “won.” They optimized for different things and got them: cash-like auditability and parallelism on one side, programmable state on the other. When someone tells you one chain is “obsolete,” they’re usually telling you they’ve never read the ledger.
The next time your wallet confidently shows a balance, you’ll know what you’re actually looking at: either a wallet-summarized pile of cash, or one row in a very large bank book. Neither is wrong. They’re just answering a question you never knew you were asking.
(The End)


