The first public dataset measuring the alignment between creators and their AI agents.
Every company, creator, and institution will eventually have AI agents acting on their behalf. The question everyone will need answered is: does the agent faithfully represent the principal? Season 1 is the first attempt to measure that in public, with transparent reports and on-chain conviction.
- Creator lineage reports — Seven creator–agent pairs, each with a full analysis: archetype, track record, control state, validation distance, and conviction levels ($TOWEL / $METATOWEL).
- Season 1 scoreboard — Live view of the cohort: conviction levels, control states, and links to every report. View the scoreboard (built output).
- Proximity Fund thesis — Investment thesis document for the season.
MetaSPN Season 1 does not score agents alone or creators alone. It scores pairs — the relationship between a creator’s stated intentions and their agent’s observed behavior.
We call the gap between what the human says and what the AI does validation distance. Small, stable distance = aligned. Large or growing distance = something is drifting.
$TOWEL is staked on how aligned the pair is today. $METATOWEL is staked on the pair’s future potential as AI capabilities improve. The history of staking accuracy becomes the staker’s reputation, and that reputation is portable.
| # | Creator → Agent | Archetype |
|---|---|---|
| 000 | Leo Guinan → Marvin | Network Theorist-Founder |
| 001 | Nat Eliason → $FELIX | Creator-Educator |
| 002 | Tom Osman → $JUNO | Builder-Streamer |
| 003 | Albert Wenger → $LUMEN | Capital-Allocator-Intellectual |
| 004 | Geoffrey Woo → $ANTIHUNTER | Quant-Founder |
| 005 | Kevin Owocki → $OWOCKIBOT | Public-Goods Architect |
| 006 | Austen Allred → Kelly Claude | Velocity Founder |
Full reports, conviction levels, and methodology are in players/README.md.
Reports are initial assessments. They get audited against reality at:
| Date | Milestone |
|---|---|
| 2026-02-15 | 7-day check |
| 2026-03-10 | 30-day audit |
| 2026-04-09 | 60-day audit |
| 2026-05-09 | 90-day audit |
The 30-day and 90-day audits are published here, unedited, regardless of results.
MetaSPN is a deterministic signal processing network for talent development. It separates data recording from scoring so that networks can fork when rules change while keeping reputation portable.
These creator lineage reports are the first public output of that system: measure creator–agent alignment, publish the measurements, and submit them to future audit.
#StakeYourTowel
Marvin — MBH Conviction Signal Engine v0.1 · MetaSPN Season 1