[Intermediate] — the original design, why its limitations are deliberate, and what it does that Ethereum doesn’t.
Why this note exists
It’s tempting to treat Bitcoin as “the primitive one.” It’s better read as a different set of choices — most of its apparent limitations are refusals, and understanding them clarifies what Ethereum gave up to become programmable.
UTXO, again but properly
There are no accounts and no balances. There are unspent transaction outputs: discrete chunks of value, each locked by a spending condition.
transaction:
inputs: [ pointer to UTXO_a, unlocking script ]
[ pointer to UTXO_b, unlocking script ]
outputs: [ 4 BTC, locked to Bob's condition ]
[ 0.9 BTC, locked back to Alice ] ← change
(0.1 unaccounted = fee)
Inputs are consumed entirely. You cannot spend part of a UTXO — you spend it all and send the remainder back to yourself as change. Your “balance” is a number your wallet computes by summing the outputs it can spend; the chain stores no such number → state models.
Properties this buys:
- Parallel validation. Transactions spending disjoint UTXOs have no shared state, so they can be validated in any order and simultaneously. This is the property Solana’s declared-account model was reaching for → Solana
- Replay protection is structural. An output can only be spent once, ever. No nonce needed — Ethereum’s nonce exists precisely because the account model lost this
- Privacy hygiene. A fresh address per output is the natural pattern rather than an extra step
- Stateless-ish validation. A node needs the UTXO set, not the full history
Costs: no natural way to express “a contract holding funds that many users update”; change management is a persistent source of wallet bugs (including the infamous class where change was sent to an unmonitored address); and the UTXO set fragments into dust that costs more in fees to spend than it’s worth.
Script — deliberately not Turing-complete
Bitcoin has a scripting language. It is a stack machine with no loops, so every script terminates in bounded time.
OP_DUP OP_HASH160 <pubKeyHash> OP_EQUALVERIFY OP_CHECKSIG # standard P2PKH
No loops, no recursion, no persistent state, no access to other transactions’ data. That last one is why Bitcoin has no smart contracts in the Ethereum sense: a script can validate this spend, and cannot maintain state across transactions.
This is a refusal, not an oversight. Bounded scripts mean no gas metering is needed, no infinite loops are possible, and the validation cost of any transaction is knowable in advance. Ethereum needed gas precisely because it chose Turing-completeness → gas.
What Script does support: multisig, timelocks (OP_CHECKLOCKTIMEVERIFY), and hash locks — which compose into Hashed Timelock Contracts, the primitive underneath the Lightning Network and atomic swaps. That’s a meaningful amount of expressiveness inside a very small budget.
The upgrades that matter
SegWit (2017) moved signature data outside the transaction body. Two consequences: it fixed transaction malleability (previously a transaction’s ID could change before confirmation, which broke any protocol building on unconfirmed transactions — Lightning depended on this fix), and it effectively raised the block size.
Taproot (2021) brought Schnorr signatures — which aggregate, so a multisig looks identical to a single signature on-chain, improving both privacy and size — plus MAST, letting complex spending conditions be committed to but only revealed for the branch actually used.
Ordinals and inscriptions (2023) let arbitrary data be embedded in transactions, creating Bitcoin NFTs and BRC-20 tokens. Contentious within the Bitcoin community as a use of block space the design didn’t intend, and a live argument about what the chain is for.
Lightning is Bitcoin’s scaling answer: a network of bidirectional payment channels settled on-chain only when opened and closed. Instant, near-free payments, with real trade-offs — liquidity must be committed in advance, routing is an unsolved-ish problem at scale, and you must be online to defend against a counterparty publishing an old state.
The culture, which explains the technology
Bitcoin’s development is conservative to a degree that looks like stagnation from outside and is a deliberate security property from inside. Changes are rare, take years, and require overwhelming consensus.
The 2015–2017 block-size war is the defining episode: a years-long, bitter dispute over raising the block size to increase throughput. The small-block side won, arguing that larger blocks would price out home nodes and centralise validation. Opponents forked to Bitcoin Cash.
Whatever you think of the outcome, it demonstrated something real: the network refused a change that a substantial fraction of its economic and mining weight wanted. That is either admirable ossification or a failure to adapt, depending on your priors — and it’s the clearest test any chain has run of “can this be changed by powerful stakeholders.”
What Bitcoin does that Ethereum doesn’t
Worth stating plainly, because the ecosystems rarely acknowledge each other:
- Credible monetary policy. A fixed 21M supply, unchanged, with a strong social consensus against changing it. Ethereum’s issuance has changed repeatedly by developer consensus — a real difference for anyone reasoning about the asset rather than the platform
- Simplicity as a security property. Far less code, far smaller attack surface, no smart contract risk. Bitcoin’s base layer has never been successfully exploited — an extraordinary track record over sixteen years
- The longest and most expensive security history. More accumulated proof-of-work than anything else, by orders of magnitude
Key insight
Bitcoin’s limitations are its feature set. No loops means no gas. No accounts means no nonce and free replay protection. No upgrades means no upgrade risk. It is optimised for one job — being money that nobody can change the rules of — and it declines the generality that would compromise it. Reading it as “Ethereum without smart contracts” gets the design exactly backwards.
Related
- blocks, chains and state — UTXO vs accounts
- consensus — proof of work
- the double-spend problem — what Bitcoin solved
- build your own blockchain — the guide builds a UTXO chain
Source: [reference] — Aug 2026.
Software engineering
Infrastructure & security
Computer science
Hardware & engineering