Current products and systems

Guideline committees, payers, and regulators make decisions based on evidence that shifts. They usually find out it moved after the decision is already stale.

Forecasts are frozen before the outcome is known and scored afterwards, so the record can be checked rather than retold.

I built the registration system, a deterministic scoring engine that freezes forecasts before the outcome is known and scores them against the public record, and the resolution rules that decide what counts as the forecast coming true.

The method and the live public ledger are both at nextconsensus.com — that ledger is the authoritative record, including how many forecasts have been filed so far.

Founder

When a source silently edits, retracts, or updates, every downstream claim built on it goes stale — and nobody notices.

Replay the source's revision history and emit every change as a deterministic event. Downstream systems decide what matters.

A graph-based claim-history engine, published to npm, with a daily observation bot running in production.

Source code, docs, and commit history at github.com/refract-org/refract.

Change detection should be deterministic. If the observation layer is probabilistic, you can't trust the rest of the stack.

Any system that cites sources — citation graphs, evidence trails, compliance records — needs to know when those sources change. Refract is the observation layer for that — built before source-level provenance became a standard requirement in AI and compliance systems.

Founder

AI systems get deployed, then drift. Nobody owns what happens when the evidence changes, the model degrades, or a patient is harmed.

Design the questions into the product: who owns the recommendation, who can override it, what happens when it's wrong, and who corrects it when the evidence moves.

Six published standards covering authorization states, reconsideration loops, escalation paths, and safety cases — mapped against NIST AI RMF, ISO/IEC 42001, and the EU AI Act.

Full standards and operating model at ethotechnics.org.

Institutions don't lack governance frameworks. They lack the product infrastructure to enforce them — the override paths, the correction loops, the audit trails that make accountability real instead of aspirational.

Every hospital, payer, and health system deploying AI faces the same governance gap. The standards are open so they can adopt without locking into a vendor — published while most institutions were still treating AI governance as a future problem.

Author

Automated decisions move faster than the institutions that are supposed to review them. The gap is where people get hurt.

Write about where accountability thins when institutions absorb automated decisions — and what keeps systems answerable.

231 essays on institutional power, automation friction, and what keeps systems answerable.

All essays at thecrumple.zone.

When a system harms someone, the question is never 'was the algorithm wrong?' It's 'who owned the decision the algorithm fed into?'

The writing informs the product work. The concepts — cognitive scarcity, governance by attrition, asymmetric irreversibility — show up in the framework and in how I build.

Author

AI capability drift and dependency decay are currently unobservable in automated agents — a gap that will widen as agent deployments scale.

Model an agent's tools, skills, and dependencies as a directed graph so you can see what's decaying before it breaks.

A SQLite-backed graph engine with 17 MCP query tools, a tracking plugin for config changes, a visualizer, and executable pruning — distributed via npm and brew.

The repository, live demo, and npm package are public at github.com/zz-plant/capability-graph.

Capability rot is silent. The first sign of a decaying skill is never 'it broke' — it's 'the agent stopped using it and nobody noticed until a downstream capability failed.'

The same observation-layer pattern I've been building since Epic — alert-fatigue monitoring for clinical decision rules, source-change detection in Refract, capability-drift tracking here. You can't fix what you can't see decaying.

Developer