Aggregation theory for credentials

Work in progress. This is a public outline; prose to follow.

The thesis

Ben Thompson’s aggregation theory says the internet flipped value capture from those who control supply to those who control demand — and the deepest form of that control is owning the discovery layer. The same dynamic is now arriving in learning and credentialing, but with a twist: the scarce resource is shifting from human attention to trustworthy, machine-readable evidence of skill.

1. Aggregation in two minutes

• The pre-internet moat: control distribution of scarce supply.
• The internet moat: own demand by owning discovery (Google, Amazon, Facebook).
• The flywheel: best experience → most users → most suppliers → better experience.

2. Why credentials are a discovery layer

• Employers don’t buy skills; they buy trustworthy signals of skills.
• Degrees used to be the discovery layer (prestige + scarcity).
• Micro-credentials break the degree monopoly — but recreate the discovery problem at higher resolution: which credential, from whom, signals what?

3. The agentic rupture, applied to learning

• Agents don’t browse; they read structured data.
• A credential that isn’t machine-readable is invisible to the agentic web.
• Compare commerce’s Shopify/UCP move: the platform that makes merchants legible to agents wins, independent of which agent wins. Credentialing will do the same.

4. What builders should do

• Treat credentials as a protocol, not a product.
• Compete on the trust stack (identity → evidence → issuance → portability), not on the certificate PDF.
• Ship machine-readable, verifiable, portable credentials (Open Badges 3.0, verifiable credentials) before the aggregator locks the discovery layer.

5. Why this is a blue ocean for small builders

• The incumbents (Credly, Instructure) sell issuance, not trust-as-protocol.
• Nobody is writing the strategy or building the open protocol for learning discovery. The gap is wide open — see building-in-the-gap.

Open questions

• Does credential discovery converge to one aggregator (winner-take-most), or does the heterogeneity of skills keep it open for protocols?
• Who owns the learner’s credential graph — the issuer, the employer, or the learner?

Draft. Prose to follow. Cross-links: learn.mneurix.dev (the trust stack), neuro.mneurix.dev (autistic-led research as a trust model).