Economic flywheel

The loop, and the single node that decides

Four nodes is usually enough to draw one. Usage generates fees. Fees fund incentives. Incentives attract the supply side, whether that is liquidity, hardware, validators or sellers. More supply makes the product better, which brings more usage. Round it goes.

Three of those arrows are free in the sense that they cost the protocol nothing extra: more usage really does produce more fees, and more supply really does improve the service. One of them costs money. Paying the supply side is a transfer, and it comes either out of what the business earned or out of newly issued tokens.

That is the entire diagnostic. Trace the funding of the incentive arrow. Everything else about the design follows from the answer, including whether the loop has a stopping condition written into it that nobody has looked at.

The layer-one version, and the condition attached to it

4Pillars Research describes the layer-one case as a security flywheel: token value supports validator incentives, validator incentives buy security, security produces developer and user confidence, confidence produces usage and fees, and fees produce token demand. Their framework attaches an explicit condition to it, which is that the loop is self-sustaining only if fee revenue rather than block subsidy comes to dominate over time.²

Keep that condition in view, because it is the same condition in every category. A chain paying for its security out of issuance has a working loop and a deadline. The loop becomes independent of the deadline at the point where fees carry the security budget, and not before.

Two-sided loops, where DePIN makes the structure obvious

Physical-infrastructure networks make the mechanics legible because both sides are visible. Peer-reviewed work on DePIN tokenomics in Frontiers in Blockchain treats the design as a pair of reinforcing loops connecting network usage, token demand and infrastructure buildout on the supply side.¹

The design problem in those networks is that the two loops run at different speeds. Hardware deployment responds in weeks to a token incentive. Demand for the service responds over quarters or years, if it responds at all. So the supply loop spins up first, on emissions, and the question is whether demand arrives before the emissions schedule runs out of room. That timing gap is the real subject of most DePIN tokenomics work, and it is a business question wearing a mechanism costume.

Two loops that were pushed rather than turning

Terra's Anchor Protocol is the reference case. Messari's coverage documents it as a DeFi flywheel substantially subsidised by protocol emissions and reserves rather than by organic revenue.³ The loop was genuine while it ran. Deposits attracted by the yield fed the narrative, the narrative attracted deposits. What it never had was a source of income large enough to pay the yield out of, which meant the reserve was the loop.

OlympusDAO is the second. The high staking yield was the mechanism the entire loop was marketed on, and in 2022 the protocol cut its annual OHM staking yield to 7.35%, a reduction reported by The Block at the time.⁴ The loop did not break. Its price of admission was simply repriced to what the protocol could afford.

Neither of these was a coding failure and neither was a fraud in the mechanism. Both were loops whose incentive arrow was paid out of a finite pool, presented as loops that funded themselves. The published numbers were true. What was missing was the funding line under the arrow.

The test we run on a flywheel slide

Take the diagram, and for each arrow write down the asset that moves and where it comes from. Arrows carrying user demand cost nothing. Arrows carrying payment cost something. Then total the paying arrows for one turn of the loop and ask what share of that total is revenue and what share is issuance.

If issuance is above zero, calculate how many turns the treasury funds at the current rate. That number is the design's honest horizon, and it belongs in the model next to the growth curve rather than in a footnote. A loop that funds forty turns and needs eight to reach fee dominance is a plan. One that funds forty and needs four hundred is a countdown with a nice diagram.

There is nothing wrong with subsidising the early turns. Almost every network does, and issuance is a legitimate way to buy the supply side before revenue exists. The failure is not subsidy. It is a subsidy nobody has priced, running against a demand curve nobody has tested, drawn as a circle so it looks like it closes.

Our position

We treat a flywheel slide as a claim about the business, not a claim about the token. The loop is the shape of the revenue model drawn as a circle, and if the revenue model does not work when you draw it as a line, the circle does not repair it.

So the work happens in the other order. Establish what the protocol earns and from whom. Establish what it costs to bring the supply side in. Only then decide which portion of that cost issuance covers and for how long, and write the exit condition down where governance can see it.

The token is infrastructure for the business. A flywheel is a description of a business that compounds, and no mechanism makes a business compound that was not going to.