Dai launched backed by over-collateralised loans, not deposits

Inherited archive entry · retained from the earlier edition. Its factual, legal and policy claims have not all been reverified for 19 September 2026. Research scope.

A credit line opened without a bank

On 18 December 2017, MakerDAO's own account of the launch states that the first MKR governance vote activated Single-Collateral Dai and that the Dai Credit System became operational that day. The idea being tested was narrow and specific: a token pegged to the US dollar that draws its backing not from a bank's reserves but from collateral locked inside a public smart contract. The project's whitepaper frames Dai as a "collateral-backed cryptocurrency" whose value is kept stable relative to the dollar by that locked collateral rather than by a custodian's promise.

How a loan created the money, and how it was destroyed

The whitepaper describes a Collateralized Debt Position, or CDP: a user locks Pooled Ether into a contract and generates newly created Dai against it, provided the value of the collateral stays above the value of the debt by a ratio that MKR holders set through a vote. Paying down the debt plus an accruing Stability Fee lets the user reclaim the locked collateral and, in the process, removes that Dai from circulation. If the ratio falls too far because the price of Ether drops, the whitepaper's liquidation process auctions the collateral to cover the debt automatically, without asking the borrower. Should an auction fail to raise enough Dai, the system dilutes and sells MKR to make up the shortfall, which is why MKR holders, not depositors, carry the tail risk of a bad loan.

What the design admitted was incomplete

The whitepaper is candid that the December 2017 launch, Single-Collateral Dai, was a first stage rather than a finished product. It supported only one form of collateral, Pooled Ether, and relied on scaffolding it calls "temporary": a Liquidity Providing Contract to backstop Dai's price, and centrally chosen Emergency Oracles empowered to trigger a shutdown until a fully decentralised alternative was ready. That alternative, Multi-Collateral Dai, arrived in November 2019, outside the scope of this record. A system's own description of its early limits is more reliable evidence than a later summary of what it eventually became.

Questions to carry forward

• Does a quoted collateralisation ratio describe the rule enforced at the moment of borrowing, or an average across a whole system of loans?
• Who absorbs the loss when a liquidation auction fails to recover a position's full debt, and how is that loss allocated among holders?
• Which parts of a mechanism labelled "temporary" at launch were later replaced, and which quietly became permanent?

The 18 December 2017 launch set a template decentralised finance has repeated since: substitute a custodian's balance sheet for a public ledger of locked collateral and a published rulebook for what happens when prices move against it. That the same over-collateralisation and auction logic still appears in the protocol's current documentation, published since a 2024 rebrand to Sky Protocol, is evidence the template proved durable. It is not evidence that durability came free of cost, and a separate record in this archive covers how the same auction mechanism performed under the stress of March 2020.