Algorithmic Stablecoins: A Historical Class, on Record

The design, as it was written

An algorithmic stablecoin holds its price with rules. The rules control supply and demand. No reserve stands behind the tokens. The common form was seigniorage. Price above target? The algorithm expanded supply. Price below? It contracted supply. The description is accurate as far as it goes. What the design lacked was anything to sell when everyone wanted out at once.

The word algorithmic was used loosely for years. This page means the narrow thing. A protocol with real collateral is not it, whatever its software does. The pure algorithmic design held nothing. Its only asset was the rule. The rule had a failure condition.

Why the rule ran one way

The rule ran one way, and the asymmetry was the design’s fate. Expanding above target was easy. The protocol minted against demand that already existed. Contracting below target was the hard direction. The protocol had to take tokens back. With no reserve, the only payment it could offer was more of its own paper. In two-token variants, that paper was the sibling token. The exchange rate between the siblings was the whole machine.

The peg held while exit demand stayed small. The design’s defenders read that as proof. The record reads it as a condition. The system worked only while fewer people left than arrived. That is a bank-run shape, built in. The run arrived the way the shape predicted.

The canonical case

Terra’s UST is the case the class will be remembered by. Its peg ran through a sibling token. One UST was always burnable for a dollar of LUNA. May 2022: heavy UST selling pushed the price under target. The escape valve did what it was built to do. It minted LUNA to absorb the exits. Under run pressure it minted trillions of LUNA. Both tokens collapsed together ^[1].

The terminal price landed near one cent. The reconstruction’s own words: little more than a penny ^[1]. Value destroyed across the ecosystem ran to tens of billions of dollars. The figure varies by measurement. This site records the range. It never averages. Feed and pair are named wherever a price appears.

Why the class is taught as history

No large-scale pure-algorithmic issuer survives in this corpus. None is asserted to exist anywhere. The honest framing is historical. This was a class. It had a canonical case. The case ended the class. A page presenting the design as live owes the reader a dated name. This site has none to give.

The lesson travels. The design failed at the exact point it promised most. The guarantee depended on the escape valve keeping pace with exits. Every surviving class in this library holds an asset somewhere. The asset is what exit demand draws on.

How to read a survivor claim

Claims that the design lives on appear now and then. Two questions check them. What does the token hold? A design with reserves belongs on another class page, whatever its marketing says. Who says it survives, and when? A claim without a dated document is a claim this site cannot carry. The bar is plain. Name the token. Name the document. Date the read.

The test cuts the other way too. A token with rules and real collateral is not algorithmic in this page’s sense. Calling it that imports the 2022 story where it does not belong. Precision is not academic here. The label carries a documented terminal case. It should not land on designs that hold assets.

Three taxonomy corrections

The old version of this page made three errors. Each gets its fix here. Error one: it split the class into seigniorage-style and collateral-backed subtypes. Collateral-backed tokens are the crypto-backed class. That is a separate design, with real locked assets. The split is retired ^[2].

Error two: it named DAI as a known algorithmic token. The same sentence called DAI collateral-backed. DAI has been crypto-backed since its 2017 launch. Its contracts are documented today ^[2]. It belongs on the other page. The correction links there.

Error three: the FRAX label. Legacy FRAX has been called algorithmic by some and collateral-backed by others. This site asserts neither. The research leaves it there, so the page leaves it there. The issuer does state one thing about the successor, frxUSD. It is backed by tokenized US Treasury funds. That is a reserve statement, quoted as the issuer’s own ^[3].

Where this page sits

The stablecoin intro maps the four classes and marks this one historical. The crypto-backed page carries the surviving collateral design the old taxonomy blurred this class into, and the fiat-backed page carries the reserve class. The risks page turns this class’s ending into its history check, one of four documented failure modes, stated without hierarchy. The stablecoins hub frames the pillar and its record on pegs that broke, and the cryptocurrency hub is this page’s home root.