Methodology / Evidence
The evidence taxonomy.
Five evidence types, each measuring a different dimension of skill credibility. Here is what they are, why they are weighted the way they are, and how each one works.
No single evidence type tells the full story of skill credibility. A person who aces assessments but never teaches may lack applied fluency. A person with many endorsements but no session history may be well-connected but unproven. SCI requires evidence breadth, not just depth.
Breadth of evidence is as important as depth. Neither alone is sufficient.
How evidence feeds into SCI
SCI
0 - 1000
Sessions
Sessions are the strongest evidence type because they prove live, interactive competence. Teaching a concept to another person requires deeper understanding than passing a test. The 40% weight reflects this primacy.
How it works
When you complete a session, the system records a completion event with duration, both participants' ratings, and the skill being practiced. Each event is immutable and timestamped. Session quality is derived from the peer rating and whether the full scheduled time was used.
Signals tracked
Assessments
Assessments verify knowledge depth on specific topics in a controlled environment. They complement sessions by testing conceptual understanding that might not surface in a teaching interaction. The 20% weight balances their value against their limitation: they measure knowledge at a point in time, not ongoing capability.
How it works
Assessments are administered through the platform with randomized question pools. Results are scored automatically and linked to the specific skill being tested. Retakes are allowed but carry diminishing weight to prevent score farming.
Signals tracked
Endorsements
Peer endorsements create a credibility network. Unlike LinkedIn endorsements, they are weighted by the endorser's own SCI. An endorsement from an Authority carries four times the weight of one from a Newcomer. This makes endorsement trading ineffective.
How it works
Users can endorse peers they have interacted with on the platform. Each endorsement specifies a skill and an optional written justification. The system verifies that the endorser has relevant interaction history with the recipient before accepting the endorsement.
Signals tracked
External Proofs
External proofs anchor skill claims to artifacts that exist outside Lemma. A GitHub repository with commit history, a published research paper, or an industry certification adds a dimension of credibility that platform-only evidence cannot provide.
How it works
Users submit proof URLs or upload documents. The system validates accessibility and authenticity where possible (e.g., checking that a GitHub repository exists and has recent activity). Human review may be required for some proof types. Accepted proofs are linked to the claimed skill permanently.
Signals tracked
Portfolio
Portfolio items provide tangible, visual evidence of applied skill. They carry the lowest weight because they are self-curated and not externally validated. However, they add a qualitative dimension that complements the quantitative signals from other evidence types.
How it works
Users upload work samples, screenshots, design files, or other artifacts directly to their Skill Passport. Each artifact is tagged with a skill and a brief description. Portfolio items are visible to anyone viewing your Passport and contribute to your SCI based on recency and quantity.
Signals tracked
Why these weights
The weights reflect our belief that demonstrated, interactive competence (sessions at 40%) is the strongest form of evidence. Structured testing (assessments at 20%) adds verified knowledge depth. Social proof (endorsements at 15%) and external validation (proofs at 15%) provide complementary credibility signals. Portfolio (10%) offers qualitative context.
These weights are versioned and may evolve as we gather more data about which evidence types best predict real-world competence. Any weight change triggers a full score recalculation across the platform, ensuring consistency.
Evidence types FAQ
External proofs must be verifiable artifacts: GitHub repositories, published papers, certifications, or portfolio links. The system validates accessibility and authenticity. Human review may be required for some proof types.
Start proving what you know.
Early access is rolling out for individuals and teams. No credit card, no PDFs — just the things you made, made visible.