Dual-token model

Why teams reach for a second token

One token is asked to do two incompatible jobs. As a currency it should be abundant, spent freely, and stable enough that a reward feels the same this week as last. As an ownership claim it should be scarce, held rather than spent, and free to appreciate. A design that makes a single asset do both ends up punishing whichever behaviour it did not optimise for.

Splitting the roles is a reasonable response. a16z crypto's taxonomy makes the same cut from the outside: an arcade token is a spendable currency with a relatively stable value inside one product, deliberately not an investment instrument, while a network token derives its value from the operation of a decentralized marketplace and is acquired by people expecting returns.¹ Two different instruments, two different regulatory postures, two different holder motivations. On paper the split is clean.

The coupling test

Here is the question that decides whether a dual-token design is real. Is there any action in the system that requires holding or spending both tokens together? If yes, the capped token's demand is downstream of the uncapped token's economy, and the separation is a diagram rather than a mechanism.

The consequence follows immediately. When the currency's issuance outruns its sinks and participants stop performing the action that consumes both, the capped token loses its sink at the same moment. It does not decouple. It falls with the thing it was supposed to be insulated from, and it does so while the whitepaper still says the two are independent.

Axie Infinity: where AXS and SLP touched

Axie ran the canonical pairing. SLP was the in-game currency earned by winning matches and spent on breeding. AXS was the governance token, described by Naavik as the place where value accrues, used for breeding, for staking and later for land.² Note that breeding appears in both descriptions. It was SLP's only meaningful sink and it was also one of AXS's main uses.

The Verge set out the same economy from the outside in April 2022: three resources, SLP, the Axies themselves and AXS, with the game producing two of the three in constantly increasing quantities.³ So the sink that was supposed to absorb SLP also minted new Axies, adding supply to a second market, and the only reason to spend AXS was the same breeding action.

When SLP earnings stopped covering what a scholar could make elsewhere, breeding stopped. That removed SLP's sink and AXS's utility in one step. A structure with two tokens behaved, under stress, like one.

STEPN ran the same structure and got the same answer

Naavik's July 2022 review of STEPN describes an explicitly parallel design: GST as the primary in-game currency with infinite supply, the equivalent of SLP, and GMT as the governance token with a finite supply of six billion, of which roughly 10% had been released at the time of writing.⁴ Runners earned by moving; sneakers were NFTs; both tokens were required across the minting and levelling mechanics.

The scale was not trivial. Citing Degame data, the same review put STEPN's May 2022 transaction volume at $35.3 billion across 632,000 monthly active wallets, ranking it first by transaction volume among comparable projects at that point.⁴ Two months later the same author was describing both STEPN and Axie as models that were falling apart.⁴ Volume was not the constraint. The coupling was.

What genuine separation requires

Three conditions, and all three have to hold. No transaction requires both tokens. The capped token has at least one demand source that does not depend on the currency's economy at all, such as fees paid by external customers or governance rights over something with independent cash flow. And the currency's price stability is enforced by a mechanism, whether an issuer faucet and buyback or a hard supply rule, rather than left to a market that will reprice it every time rewards change.

That last condition is where the arcade token structure earns its place. If the currency is bounded by a published faucet price and an optional buyback, the design team can retune reward rates without triggering a market event, and the capped token is no longer exposed to the currency's order book.¹ Give that up and you have two tickers on one economy.

Our position on this is narrow and firm. We do not accept a two-token structure in a design review until the team has written down the specific action that would decouple them and shown that no mechanic requires both. The label is free. The separation is not.