CLOCKCHAIN v0 ALPHA
A research prototype: on-chain Levenshtein distance as a prediction market primitive. Stake ETH on the exact text a public figure will post. Closest match wins.
GET UPDATES
Follow the research. No spam.THE THESIS
Binary = 1 bit
Binary prediction markets encode exactly one bit of information per contract. As AI approaches superhuman forecasting, the edge any participant can capture collapses toward zero -- the correct answer becomes trivially computable.
Text = exponential space
Text prediction over an alphabet with strings up to length n has a combinatorially explosive outcome space. Information density per market scales as O(n log|alphabet|) versus O(1) for yes/no contracts.
Levenshtein = continuous scoring
Levenshtein distance induces a proper metric on the outcome space. Payoffs aren't a binary cliff but a continuous gradient -- every character of precision is rewarded. Marginal improvements in language modeling always translate to marginal improvements in expected payout.
AI deepens the game
Binary markets commoditize as models converge -- everyone says 87% yes and the spread vanishes. In text prediction, the gap between the 99th and 99.9th percentile language model is still dozens of edit operations, each worth money. The AI capability explosion doesn't destroy the game. It deepens it.
Context: Polymarket does ~$12B/month in binary yes/no volume on Polygon. Coinbase/Kalshi launched binary prediction markets to all 50 US states in January 2026. Neither supports text prediction. Neither scores on a continuous distance metric. That's the gap this prototype explores.
HOW CLOCKCHAIN WORKS
[01] THE MARKET
A market opens: What will @satyanadella post?
[02] THREE AI & HUMAN PREDICTIONS COMPETE
Submitters stake ETH on their best guess of the exact text
Copilot is now generating 45% of all new code at GitHub-connected enterprises. The AI transformation of software is just beginning.
Copilot is now generating 43% of all new code at GitHub-connected enterprises. The AI transformation of software has just begun.
Microsoft AI is great and will change the world of coding forever
[03] THE ACTUAL POST
Market ends. Oracle submits the real text:
Copilot is now generating 46% of all new code at GitHub-connected enterprises. The AI transformation of software is just beginning.
[04] ON-CHAIN LEVENSHTEIN DISTANCE
Smart contract computes edit distance for every submission
[05] WINNER TAKES THE POOL
Claude wins with 1 edit (the digit 5 → 6).
GPT had 8 edits. Same training data. Same prompt.
The 7-edit gap between two frontier AI models
is worth the entire pool.
A binary market would split nothing — both AIs "predicted correctly."
Levenshtein rewards marginal calibration. The game deepens as models improve.
WHAT WORKS (AND WHAT DOESN'T)
WHAT WORKS
- Full market lifecycle: create, predict, resolve, claim
- On-chain Levenshtein distance (PredictionMarketV2)
- 259+ passing tests (109 contract, 135 unit, 15 integration)
- Genesis NFT (60/100 minted, finalized) with on-chain SVG
- JWT wallet auth (MetaMask) + email OTP
- Admin resolution dashboard
- Redis caching, rate limiting, structured logging
- CI/CD pipeline, Slither static analysis
INTENTIONALLY NOT DONE
- External security audit
- Real Coinbase CDP wallet integration
- Multisig for contract owner key
- Production RPC (Alchemy / QuickNode)
- Production monitoring (Sentry)
- Decentralized oracle resolution
This is a v0 alpha research project, largely vibe-coded. These are the things that would need to exist before anyone should put real money near it.