Pre-launch · contract not deployed

Bounties for proofs
a machine can't argue with.

Lock a bounty against a statement. Anyone submits a proof. Lean's kernel checks it and either accepts or doesn't. No oracle, no committee, no vote — the answer is computable by anyone with a laptop, and every honest checker gets the same one.

Litepaper
CONTRACT not deployed · no presale · no whitelist — anyone selling now is a scam
Chain
Robinhood Chain
Supply
TBA
Presale / team alloc
None
Verification
Lean 4 kernel
01 Why this works on-chain the whole thesis in three boxes

A The asymmetry

Hard to produce, trivial to check. A proof can cost months of research or thousands of GPU-hours. Verifying it costs a laptop a few seconds. Same shape as proof-of-work — except the expensive artifact is worth something afterwards.

B No oracle problem

Everything else chains adjudicate is fuzzy — a price, an outcome, a judgement. This isn't. Two honest verifiers can't disagree, because they aren't forming an opinion. They're running a kernel.

C Somebody pays

Protocols holding tokenized RWAs. "We looked hard and found nothing" is what an audit sells. "No exploit exists" is what a proof sells. Only one is provable, and nobody is selling it yet.

02 How a bounty settles post → solve → check → pay
1

PostLock $ARISTOTLE against a Lean theorem statement. The statement is the spec and it's frozen at post time.

2

SolveResearchers, prover models, or people driving prover models submit proof terms against it.

3

CheckStatement elaborated in a clean module the solver can't touch. Proof typechecked. Axiom dependencies audited.

4

PayFirst submission that survives all three takes the pot. Artifact is public and re-verifiable forever.

03 The audit is the product try the third one — it typechecks and still fails
bounty #0001 · submission

      
      
04 What the token actually does two of three are load-bearing

Bounty escrow soft

Prizes locked in $ARISTOTLE for the bounty's duration. Honest caveat: a poster could denominate in stables instead, so this sink alone would be decoration.

Verifier bonds mandatory

Attesting requires stake. Sign off on something that doesn't typecheck and the bond goes to whoever proves you wrong. You can't verify without holding it, and that can't be routed around.

Settlement fee mandatory

Every settled bounty pays a protocol fee, routed to buy-and-burn and to verifier stakers. Scales with real throughput, not with narrative.

05 Verification roadmap Lean's kernel is too big for the EVM — for now
phase 1

Optimistic

Bonded verifiers attest on-chain with a challenge window. Anyone may re-run the kernel locally and dispute. Unlike most optimistic systems the disputed question is objective and cheap to recompute — a liar can't recruit honest confusion, because every independent checker gets the same answer.

phase 2

Zero-knowledge

Lean's trusted kernel is ~10k lines. Running it inside a zkVM to prove this term typechecks against this statement is engineering, not research. Verification becomes fully on-chain and the trust assumption disappears.

06 What could go wrong stated up front, not buried
  • The market is thin today. Demand for formal verification is real but small. This is a bet that it grows, not a claim that it already arrived.
  • Phase 1 rests on watchers. Optimistic verification is only as strong as the set of people checking — thinnest at launch, which is exactly when someone tries to farm a bad attestation.
  • Statements are the attack surface. A bounty settles against a Lean signature, and a careless signature can be satisfied vacuously. Reviewing statements matters more than reviewing proofs.
  • Provers have a ceiling. Post problems beyond current capability and the board stalls and looks dead.