Reviewed and updated: October 19, 2026.
The development of decentralized exchanges is a story about coordinating orders, capital and settlement. It is not a simple progression from unsafe exchanges to risk-free contracts. Different designs move responsibilities between users, market makers, contracts and infrastructure providers.
Separate finding a trade from settling it
An order book records proposed buys and sells. An automated market maker, or AMM, calculates trades against liquidity under a pricing rule. Either description leaves further questions: where are orders communicated, who can choose their sequence, which assets are held in contracts and how is the final transfer validated?
Off-chain order discovery can coexist with on-chain settlement. Conversely, a trade shown on a website may depend on a centralized matching service even if the user signs the eventual transaction. The interface label DEX is therefore less informative than the actual custody and execution path.
A short, verifiable Uniswap timeline
| Date | Documented event | Why it matters here |
|---|---|---|
| November 2018 | Uniswap v1 launch, recalled in the project's v3 announcement | An early deployed AMM design became a basis for later versions |
| May 18, 2020 | Uniswap announced v2 deployment to Ethereum mainnet | Its changes included direct ERC-20/ERC-20 pairs |
| September 16, 2020 | The UNI announcement was published | Protocol governance and token distribution became a separate part of the story |
These dates do not establish the first exchange of every kind, and they do not prove that a token launch caused all subsequent growth. The earlier article's unsupported exchange-origin story and broad 2020 market total have been removed. Product announcements are useful evidence of those products, not independent measurements of the entire DeFi market.
How a constant-product pool prices a trade
The Uniswap v2 whitepaper describes a constant-product design. For a simplified, fee-free illustration, take a pool with 100 units of token X and 10,000 units of quote asset Q. Its product is 1,000,000. Adding 1,000 Q leaves 11,000 Q and requires approximately 90.9091 X to maintain that product. The buyer receives approximately 9.0909 X.
The average price is 1,000 / 9.0909, approximately 110 Q per X, although the initial reserve ratio was 100 Q per X. The trade itself moved the price. Actual protocol fees, rounding, routing and other transactions change the result; these invented reserves are not a live quote.
Why more liquidity does not remove every cost
Repeat the same hypothetical trade against a pool ten times larger: 1,000 X and 100,000 Q. After adding 1,000 Q, the buyer receives about 9.9010 X, an average of 101 Q per X. More depth reduced the price impact for that trade size. It did not eliminate network fees, contract risk or the possibility that the reference market price changes before execution.
A liquidity provider is exposed to the pool's changing asset composition. Fees are one component of the result, not a guaranteed profit. A pool can attract deposits through rewards without creating durable trading demand, which is why liquidity campaigns deserve a separate analysis.
Read an exchange design through five questions
- Custody: who can move deposited assets or change the contracts controlling them?
- Pricing: is the quote from orders, a pool curve, an auction or a combination?
- Ordering: who sees and sequences the transaction before settlement?
- Protection: what minimum output, price limit or deadline does the user authorize?
- Exit: can the user withdraw the resulting asset on the intended network?
This framework is useful across generations of DEX design. It avoids treating one successful swap, an audit badge or an attractive reward rate as proof of every other property. The durable advance is programmable settlement and new market structures; evaluating a particular venue still requires its actual rules and deployment.
Sources and scope
- Uniswap: v3 announcement and v1 launch history
- Uniswap: v2 mainnet launch, May 18, 2020
- Uniswap v2 whitepaper
- Uniswap: Introducing UNI, September 16, 2020
Continue reading: Sushi liquidity migration · Comparing CEX and DEX trade routes
Editorial update: Rebuilt the chronology from dated project sources; removed unsupported origin stories and TVL totals; added two constant-product examples and an execution-design checklist.
Pause and explain
Why does the same 1,000 Q trade receive fewer X tokens in the smaller example pool?
Compare your explanation
The trade is larger relative to the pool. Under the illustrated constant-product rule it changes the reserve ratio more, so its average price is worse. This example leaves out fees and other execution effects.
Trading is one use for pooled assets. Lending introduces borrowing, collateral and withdrawal conditions.
All five lessons · Change your learning level
Learning sequence added October 20, 2026. The article’s factual review date is shown above.