Four types of proof. Each weighted differently.
Not all evidence is equal. Here's exactly what Lemma accepts, how it's weighted, and how it gets onto your passport.
Four types of proof. Each weighted differently.
Peer Sessions
Structured, time-boxed interactions between two practitioners. Both confirm the session. Both rate each other. The record is immutable.
What counts toward weight
- Teaching sessions (you as instructor) — highest weight
- Learning sessions (you as learner) — medium weight
- Session rating and completion confirmation
Assessments
Adaptive evaluations that calibrate in real time. Pass three in a row and the bar rises. Score and difficulty level are both recorded.
What counts toward weight
- Difficulty level at time of assessment
- Score relative to the difficulty band
- Number of attempts (penalized for multiple tries)
Peer Endorsements
Formal acknowledgments from verified practitioners. Endorsements from higher-tier practitioners carry more weight than those from lower-tier ones.
What counts toward weight
- Endorser's own SCI tier (higher endorser = more weight)
- Whether the endorser has taught the same skill
- Recency of the endorsement
Working Artifacts
Files or links submitted and reviewed by a session partner. Not independently verifiable — weight is lower but still counts toward Proof.
What counts toward weight
- Reviewed by the session partner
- Linked to a specific session record
- File type and completeness
Why evidence diversity matters.
- No single event type dominates the score
- Teaching requires different evidence than learning
- Freshness decay prevents coasting on old proof
- Peer endorsers carry reputational skin in the game
Questions about evidence types.
Teaching a skill requires a depth of understanding that simply practicing it does not. If you can explain a concept clearly to someone who doesn't know it, and they come away with better understanding, that is strong evidence you know the material. It's also much harder to fake.
Partially. Linked repositories are analyzed for activity and code quality signals, but carry lower weight than peer-validated evidence because they can't be contextualized without a session partner. They're supplemental, not primary.
Not retroactively. Evidence events are created at the time of the interaction. This prevents backdating and keeps the audit trail trustworthy. If you have past work, the way to evidence it is to have a peer session where you demonstrate it.
LinkedIn endorsements carry no accountability. You can endorse someone you've never worked with. They are traded as social favors. Lemma requires evidence that involves at least one other verified practitioner who has skin in the game.
Your evidence is immutable from the moment it's created.
Start a session today. It goes on your record permanently.