Flop Labs' Token Math Doesn't Compile — And That's the Least of It

Flop Labs' second-phase tokenomics draft fails to reconcile: allocations sum to 182 billion against a stated 181 billion, and team share reads 10.8% in prose versus 11.0% in the table. The project positions as a Proof of Useful Inference (PoUI) plus AI Agent economy, with miners taking 48.6% of main allocation — a DePIN-flavored compute network, not a pure software protocol. The draft stresses no VCs, no presale, contribution-only, yet discloses no testnet timeline, TPS, latency, slashing conditions, or audit. The structural defect: miners appear twice (48.6% main, 6.6% airdrop) and validators twice (6.5%, 6.6%), with no specification whether these are additive or overlapping — if additive, miners control roughly 55.2%. The 24.3% airdrop has no vesting curve, cliff, or release schedule. Supply runs halving-driven inflation into a permanent 0.5% annual tail, no hard cap — tame by DePIN standards, but without burn mechanisms or compulsory FLOP fee settlement it is simply slower dilution. No paying customer is named; if rewards come purely from emission, the model becomes subsidy-driven compute sputtering. Team plus foundation at 10.8% sits below industry norms, lowering insider-dump motive but not delivery risk. The 'no VC' banner cuts both ways: principled fair launch or absent institutional diligence, thin liquidity, and amplified post-TGE volatility. The deeper blind spot is verification: PoUI collapses if inference correctness cannot be checked cheaply and trustlessly, yet no challenge period, fraud proof, verifier sampling, or quantified slashing math is disclosed. Three items must resolve before any position: the airdrop unlock schedule, reconciliation of the miner/validator double-count, and the actual mechanism by which useful inference is verified and paid for from outside the emission. Until those compile, the rest of the table is decoration.