experiments

External Forma Value Validation Pilot

Compare Forma-assisted and disciplined Markdown workflows for representative external knowledge-maintenance work without treating qualitative case narratives as measured outcomes.

Hypothesis

For selected recurring knowledge work, Forma makes structure, evidence, review boundaries, and reuse more reliable than a disciplined Markdown baseline without imposing an unsuitable domain model.

Case Corpus

The pilot starts from two published, user-authorized qualitative narratives in choral-io/knowledge-research:

They identify candidate domains and recurring jobs. They are not experimental results.

Protocol

  1. Recruit three to five external teams or maintainers across at least two domains.
  2. Select one representative recurring knowledge job and bounded corpus per participant.
  3. Record the disciplined Markdown baseline first.
  4. Configure a minimal Forma workspace that uses the participant's own vocabulary, templates, views, and review boundaries.
  5. Repeat the same job and collect the metric definitions without product telemetry.
  6. Record a per-claim decision: continue, narrow, reposition, or stop.

Stop Conditions

  • The corpus is too sensitive or consent is incomplete.
  • The compared jobs are not equivalent enough to support interpretation.
  • Setup burden dominates the maintained-work benefit for the selected team.
  • Evidence does not support a clear next product claim or scope.