Labs contains investigations that test a question through experiments and evidence. Northstar, the first published investigation, uses a fictional scenario and synthetic tests; its findings and limitations are set out below.
Can authority checked earlier be relied on when an irreversible action actually occurs? A fictional address-change scenario tests the gap between a valid check and a committed effect.
LEARNING / RESEARCH POC. NOT PRODUCTION APPROVED.
The combined authority experiment retained six failing assertions alongside 39 passing ones. The failures are part of the finding.
Inspect the investigationLabs records experiments and their findings without disguising uncertainty. Related applied work may reference an experiment’s findings; the experiment’s full account stays here. Each published experiment will show the problem, hypothesis, architecture, experiment, evidence, learning and next decision.
An experiment may progress, change direction or conclude that its hypothesis was not supported. All three outcomes can produce useful knowledge.
These stages describe the activity and maturity of an experiment. Evidence and limitations are stated separately. Validation means testing is underway, not that the work is validated. Not every experiment should become a product.
The problem and context are being investigated.
System boundaries, feasibility, risks and controls are being shaped.
A focused build is testing a consequential assumption.
Evidence is being tested beyond the initial prototype.
The work has earned a transition into product engineering.
Work has stopped, with the reason and learning retained.
