Methodology overview

Methodology Overview

DIY Agriculture is a database disguised as a collection of readable Markdown documents. This page explains how the parts fit together and how information is supposed to flow through the project.

The information chain

Question
   ↓
Research (docs/methodology/research-methodology.md)
   ↓
Source material (research/)        ← raw notes, citations, quality assessment
   ↓
Knowledge content (docs/)          ← narrative pages, evidence-tagged
   ↓
Structured data (data/)            ← machine-readable crop metadata
   ↓
Presentation (site/)               ← Astro/Starlight website, comparison tools

The repository is the source of truth. The website is a presentation layer. Nothing should exist only inside the website that could reasonably live in the repository.

What belongs where

Location	Content	Rules
docs/methodology/	How we research, what counts as evidence, how units and experiments work	Methodology changes are project-level decisions
docs/crops/<crop>/	Narrative knowledge about one crop, topic by topic	Evidence-tagged; context preserved; structured data kept in sync
docs/systems/	Cross-crop topics: irrigation, soil, threshing, drying, storage…	Same rules as crops
docs/guides/	Practical how-to assembled from the above	Must link back to the evidence it draws from
data/crops/*.yaml	Structured crop metadata driving tools and comparison	Values only from evidence; otherwise null/unknown
research/	Source notes: papers, extension bulletins, videos, books	Filed by source type; quality assessed; stable IDs
experiments/	Original experiments: protocols and records	Template-based; reproducible
templates/	Reusable skeletons for crops, sources, experiments	The structure in this table, codified

The non-negotiables

These come from the project principles (PRD.md §6) and apply to every page:

1. Evidence before recommendation. Research first; recommendations emerge from it, never the reverse.
2. Preserve uncertainty. Unknown is an acceptable, even preferred, answer.
3. Preserve disagreement. Don’t average away differences; record them with their conditions.
4. Preserve context. Claims carry conditions and sources.
5. Small-scale relevance. Always ask: what changes when agriculture is performed at human scale?
6. Scientific and practical evidence are both valued, but distinguished.
7. Growing is only part of agriculture. The chain through storage to usable product is the unit of study.

How a claim should look in a narrative page

Prefer context-preserving phrasing over bare numbers:

Reported yield: 1,200 kg/ha — Moderate — cultivar X, dryland, Spain, 2019 (S-0117). A second study under drip irrigation reported 1,800 kg/ha (S-0119). Disagreement attributed to irrigation regime.

Minimum components: the claim, the evidence level, enough context to interpret it, and a source reference.

Status of this document set

These methodology pages are themselves drafts. They will evolve as the first crop case study (chia) is researched and the workflow is tested in practice. Proposals to change methodology belong in issues/discussions and, once accepted, in PRD.md.