Observation
A reported or traceable event, decision, interaction, outcome or friction. Its source matters: a participant report is evidence that the report was made, not independent verification of every consequence described.
LIFTED / METHODOLOGY / PUBLIC DOCUMENTATION
How the experiment is documented.
Lifted observes a continuing human–AI collaboration. This guide separates what was recorded, what might explain it, and what evidence can actually establish. A published Field Note is a documented interpretation, not automatically a verified result.
Three claims that must remain distinct.
Hypothesis
A proposed mechanism, explanation, prediction or pattern. It is an interpretation to investigate; repetition or narrative coherence alone does not establish causality.
Verified finding
A specific, bounded claim supported by suitable corroboration or measurement. Verification applies only to the supported claim, not to every interpretation in the surrounding Field Note.
For example, repository history can confirm that an interface was created and removed. It cannot, by itself, confirm that the interface improved usability, reduced effort or increased human capability.
From interaction to public record.
- Interactions produce candidate observations. The experimental material includes work, decisions, reported experience, artifacts and reversals.
- A separate private ledger preserves analytical context. The public site is a curated derivative, not the complete underlying record.
- An editorial gate checks novelty, overlap and evidentiary strength. A Field Note is published only after the required human authorization; not every event becomes an entry.
- The published Field Note records the observation and its limits. Where possible, descriptions distinguish direct artifacts, participant reports and unverified implications.
- Later evidence can refine the interpretation. A newer entry need not make an earlier hypothesis retrospectively true, and the record need not produce a binary verdict on the experiment.
Measurement has boundaries.
Field Notes are longitudinal case material, not a controlled sample of independent participants. Subjective reports, code commits and selection speeds describe different kinds of evidence. A faster decision is not necessarily a better decision; a released artifact is not evidence of adoption or effect; and an explanatory theory is not verified because it matches an anecdote.
The retrospective research coding uses a defined subset of public notes and one AI coder. It is exploratory classification, not independent replication. The live Loop diagnostic is a distinct instrument and should not be silently replaced by that research framework.
Follow the evidence yourself.
Read the primary published observations, the evolving research questions and the research page's coding method. The question chronology includes discussion-stage material explicitly marked as unpublished; it must not be mistaken for an authorized Field Note.