Everything On-Chain

The phrase "smart contract" predates Bitcoin by fifteen years. In 1994 Nick Szabo proposed that contractual terms could be written as computer protocols and executed automatically by machines, rather than relying on a court to make things right afterwards. In a 1996 follow-up essay he gave the vending machine as an example: insert a coin and delivery follows, with no need to trust the vendor. Some twenty years later, the idea grew into a programmable machine on Ethereum.

In late 2013 a nineteen-year-old Vitalik Buterin posed a question: if a blockchain can keep accounts, why can it not run arbitrary programs? His answer was to build a Turing-complete virtual machine into the chain itself, so that any application could be deployed as a smart contract. It proposed turning the blockchain from a ledger into a general-purpose "world computer".

Gavin Wood’s yellow paper specified every rule of the Ethereum Virtual Machine formally: each opcode, each unit of gas, each state transition. With it, any team can write a client that behaves identically.

On July 30, 2015, Ethereum’s genesis block was created and the Frontier phase began — the paper became a running network. Alongside it, Solidity became the principal language for deploying contracts, a language designed for the premise that code is the contract. From that day, anyone could install a program into this world computer.

Three short technical standards determined the shape of on-chain assets. What they share is a common interface: follow the same function names and any wallet, exchange or marketplace will recognize what you issue automatically. Interoperability here was not negotiated commercially — it grew out of open standards.

On September 15, 2022, Ethereum completed "the Merge": mining (proof of work) was retired, and validators who stake ETH as a deposit took turns producing blocks instead (proof of stake). Energy use fell by roughly 99.95%. It is often described as one of the most complex live engineering migrations in blockchain history: a network worth hundreds of billions of dollars changed its consensus mechanism without going offline. Which mechanism is more secure is still debated (see the Open Questions Wall in Culture & the Future).

In August 2018, people in a Telegram group were trying to name the Ethereum open-finance movement. Candidates included Open Horizon; according to people involved, the winner was "DeFi" (pronounced DEFY). Those present included Inje Yeo (Set Protocol), Blake Henderson (0x) and Brendan Forster (Dharma). A term that would come to describe tens of billions of dollars was coined in a chat.

Traditional exchanges match buyers and sellers through an order book. On-chain, matching is too slow and too expensive. In 2016 Vitalik asked on Reddit whether prices could instead be set automatically by a formula, the way prediction markets run. Two years later Hayden Adams built that idea into Uniswap: liquidity providers deposit two tokens into a pool, traders swap directly against it, and the constant product x·y=k sets the price from the amounts in the pool.

Banks do four things: lend, settle, price and liquidate. DeFi rewrote each as a publicly readable contract. The four exhibits below are the key pieces of that rewrite. A public contract can be audited — and its flaws are equally public. The next section looks at the known risks of this design.

The previous section introduced DeFi’s building blocks; this one looks at its known risks. The four below are not one-off accidents, but costs these mechanisms carry by design.

A decentralized exchange has no order book — price is set automatically by the constant-product formula x·y=k. Drag the slider to sell ETH and watch price slide with trade size (slippage).

Every action on Ethereum pays a gas fee whose price floats with congestion. Drag network busyness to see gas price and transfer cost change.

Before the word NFT existed, people were already asking whether a particular amount of bitcoin could be "colored" so as to represent a real-world asset. Meni Rosenfeld’s 2012 colored-coins paper set out the idea systematically — the theoretical root of on-chain assets and NFTs, and a forerunner of later Bitcoin-based asset experiments such as Counterparty and the Rare Pepes (see Culture & the Future).

On May 3, 2014, the artist Kevin McCoy minted a work titled Quantum on Namecoin, later reminting it on Ethereum. It is often regarded as one of the earliest NFTs; which work counts as “the first” is disputed. In June 2021 it sold for about US$1.47 million in Sotheby’s "Natively Digital" sale.

Between 2017 and 2021 a handful of collections defined the culture of NFTs: the avatar as identity, the algorithm as authorship, holding as membership. Each item below carries its on-chain contract address. (You can check for yourself that each one exists and is unique.)

In May 2014 Vitalik wrote "DAOs, DACs, DAs and More: An Incomplete Terminology Guide", the first systematic attempt to draw terminological boundaries around decentralized autonomous entities. If smart contracts let code execute an agreement, DAOs ask something further: can an organization exist without a company and without management?

From an idea to tens of thousands of people jointly controlling one wallet, DAOs ran three landmark experiments in under a decade. One ended in disaster, one became a widely copied governance template, and one brought the word into public view.

The mainnet sacrifices throughput for security and decentralization — not a defect but a deliberate trade: if everyone on Earth is to verify the entire history independently on an ordinary computer, blocks cannot be large or frequent. The price is expensive gas and low throughput when the network is busy (you can feel this in the gas exhibit above). Layer 2's approach is not to loosen the mainnet’s limits, but to move computation off-chain and return only the result and a proof.

All rollups move transactions off-chain, compress them, and hand the result back to the mainnet. What differs is the reason to believe the off-chain computation was correct: optimistic rollups assume it was and leave a challenge window in which fraud can be proven; zero-knowledge rollups prove correctness outright with cryptography. There is also a different road entirely — not scaling one chain, but connecting many.