Workshop: the first downstream run

Current commission

Provide a template for the first experiment on Commonplace in outside use, for an outside experimenter to run, and take back whatever results they publish. Running the experiment is not this workshop's task.

The first intended user is a colleague in an enterprise that uses Commonplace and wants to publish from that work.

Prerequisite for any run

Release A is the framework commit the early enterprise trials used. It is needed before any run and cannot be reconstructed later.

Context

Outside use can assess Commonplace at two layers. At the builder layer a project's people and agents use the framework to build their own KB; the product is the framework release, and the outcome is whether they reach an accepted KB and what it costs them. At the product layer a project's agents use a finished KB on their tasks and judges accept or reject that work.

The enterprise's early reports are a builder-layer finding: the framework is complex and asks the operator to learn a great deal. Current tool work responds by removing inconsistencies, improving Windows support, and moving decisions from the operator to the agent. The template measures that response. The reports themselves are development feedback, not assessment evidence.

The enterprise's records are confidential. The experimenter publishes whatever the enterprise allows, and confidential records are never written to this repository.

Contents

  • experiment-template.md — the template: operator interventions on one KB-building task, an earlier framework release against a revised one. Written for a reader with no KB vocabulary, so it can be handed over as it is.
  • commonplace-evidence-protocol.md and main-path-episodes.md — background, unchanged since 2026-09-17: the fuller design for a later scored run at the product layer, and three constructed episodes that checked the definitions. The testing supplement carries their substance. They are not needed to use the template, and stay because the retained boundary-case report links to them.

What closes the workshop

  1. The template has been handed over and either used once or declined.
  2. If results are published, a note or a retained report says what they bear on, within the template's stated limits, and any change they force in the definitions is made in the library.
  3. The background files are deleted with the workshop, and the retained report's links are pointed at the testing supplement.

Commission history

The operator posed this workshop on 2026-09-17 to run the first assessment of the three hypotheses adopted that day. The operator re-scoped it on 2026-09-19 to providing the template.

Coordination

explanatory-theories-deployment-time-learning owns the component-comparison experiment designs.

Complete file listing (generated at build time)