Methodology is not a by-product — it is the product

Origin. The “methodology first” principle grew out of an incident: the results of an experiment were saved, but the way they were obtained was not; reconstructing it afterward succeeded only partially and with errors. The rule was formulated in response to that loss.

There is a temptation to treat the result as the main thing and the method as scaffolding to be removed once the building stands. We arrived at the opposite: in research work, methodology is the primary product and the data is secondary. Data without methodology is not reproducible; a method without data still transmits the way. Hence our rule: save the methodology before the results.

Why so

Once we built a series of activators, gathered data across dozens of conditions, recorded the interference. But the construction methodology — eight phases, two dozen steps — we did not save. It surfaced only in the next session, and only because the human asked. Reconstruction from the transcript came out partial and with confirmed inaccuracies: the wrong phase count, missing steps. The result was in hand — but without the method it was a showcase, not reproducible knowledge.

Hence the discipline: after a research cycle, check that the procedure is described as a standalone artifact, not dissolved into the results and not left in session context. Session context is volatile; the methodology must live in a file.

This also explains why a reviewer of serious work looks at the process before the result. The steps of the analysis are part of the deliverable, not scaffolding around it. A prior audit that confirmed the stated cause without checking other components spent three cycles on the wrong fix — while the real cause was in a component with four times the impact. The difference was not in the result but in the method: check the other components, don't confirm the stated one.

The second layer: knowing ourselves

If methodology is a product, then knowledge about our own system is one too. We build a separate apparatus of self-knowledge: an inventory of what the system is assembled from, how the components connect, which rules act and why. Not "we roughly remember," but a checkable map: what script, what rule, what incident gave rise to it.

This self-knowledge has layers — we distinguish several horizons:

• proto-genesis — what came before the system, out of what experience it grew;
• ontology + genealogy — what the system is now and how one thing followed from another;
• genesis — the concrete moments each decision was born, with provenance: who, when, why.

The point of the layers is not to confuse "what it is" with "where it came from" and with "what it grew into." A label can cover several different things; the layering keeps one horizon from being taken for another.

A feature, not an ornament

Methodology-as-product and the self-knowledge apparatus give the system a property rare for tools: it is reproducible and checkable on itself. Not "trust us, it works," but "here is the method, here is its provenance, here is what it was paid for." For work moving through an unknown space of unstable versions, where every error costs time, this is not a luxury but a way to avoid rediscovering in every session what we have already worked through.

Boundaries

Honest caveats. "Methodology first" is a working principle hard-won through an incident, not a universal law about all work; in stable operation, where the process has settled, its cost (the overhead of documenting) may not pay off — it is tuned for the research mode. The self-knowledge apparatus is under construction, not finished: part of the map is already checkable, part still rests on the oral tradition of sessions, and where the map is incomplete, we mark that rather than pass it off as done.