Skip to content

Source quality

Not all sources are equal, and this project deliberately mixes research-grade and practical evidence. This page defines how a source is assessed, where it is filed, and how it is cited, so that “worth keeping” and “credible” stay separate judgments.

Every filed source gets scored on these dimensions:

  1. Rigor — Is the method visible? Controlled and replicated, or a description?
  2. Relevance to humans scale — Is the practice transferable to a garden or small plot, or does it assume machinery/hectare scale?
  3. Specificity — Does it address the exact crop and question, or a distant cousin?
  4. Transparency — Are conditions, cultivars, dates, and methods stated?
  5. Recency — Is it current enough for climate/cultivar/agronomy claims? (Some classics stay relevant forever; label their age.)
  6. Independence — Is there a commercial or advocacy interest that could motivate the claim?
  7. Corroboration — Do other sources independently agree?

These feed the evidence level (evidence.md) of the claims it supports, but they are not the same thing: a high-quality source can still report a single Limited-level observation; a low-quality source can be the only report of a real technique.

Type Directory Default assessment
Academic paper research/academic/ Assess rigor rigorously; note framing
Extension/agronomy bulletin research/extension/ Note region
Government research/government/ Note jurisdiction
Book research/books/ Note edition/age
Commercial farming info (see below) Flag commercial interest
Practical grower / community research/growers/ Default lower evidence
Video (YouTube etc.) research/youtube/ Capture timestamped claims

Commercial material has no dedicated folder in the initial structure; file it under the most relevant type and flag the commercial interest in the source note.

Every source gets a file in research/<type>/ following the templates:

Naming: research/<type>/<first-author>-<year>-<short-topic>.md.

Assign each source a stable id S-####. This id is what crop pages and structured data files reference, never the filename. Allocation rules:

  • The first contributor to identify a source… is not the owner; there is no ownership. Any contributor may assign a fresh ID from the lowest unused number.
  • IDs are chronological/batch, not grouped by crop (a single source can cover many crops).
  • Keep research/ searchable; the source id is final once cited anywhere.

Inline, compact:

M · scientific — cultivar X, dryland, Spain, 2019 (S-0117)

And the full detail lives in the source note. The source note must record a bibliographic reference (journal, volume/pages/DOI or URL, authors, year) so any reader can locate the original.

Distinguish three things in a source note:

  1. What the source claims (verbatim/close paraphrase, in its context).
  2. What we assess about its quality/credibility.
  3. What evidence level we assign the resulting claim.

Keeping these in separate lines prevents us from laundering a source’s opinion into our own.