Skip to content

Source Selection and Refresh

Not every upstream dataset, paper, or supplement becomes public evidence the moment it is found. This page explains how the repository decides what to track, what to deepen, and when a refresh changes the public story rather than only the counts.

What This Page Helps Explain

  • why some source families are tracked deeply while others stay thinner
  • why a refresh can change public wording instead of only changing counts
  • why the repository sometimes keeps context visible while refusing a stronger sample-backed claim
  • how selection, refresh, and review pressure fit together

How Sources Are Chosen

Source selection is the decision about which inputs are worth carrying into the tracked system at all. That decision depends on fit, coverage, recoverability, and whether the material can honestly support the kinds of public claims this repository wants to make.

For example:

  • a pollen dataset may be valuable because it gives regional environmental context
  • an archive project may be valuable because its samples matter historically
  • a paper may still be too thin for public animal mapping if its supplement does not support recoverable sample rows

Selection matters because this repository is not trying to collect everything indiscriminately. It is trying to build an evidence system that can explain why one source deserves public attention while another remains background, partial, or blocked.

What A Refresh Can Change

Refresh is not just a mechanical re-download. In this repository, a refresh is an evidence event. It can change counts, improve locality support, expose new blockers, or force public outputs to narrow their wording.

That is why refresh work is paired with:

  • contracts that say what a source family should publish
  • review surfaces that say what changed
  • release gates that prevent public overstatement

Why Uneven Refresh Matters

Refresh work also creates unevenness. Some source families move faster than others, and that changes what the repository is allowed to say about overall coverage. When the weaker parts of the system have not caught up, broader confidence has to stay limited. That refusal is part of the honesty model, not an embarrassment to hide.

Use This Page When You Need To Know

This page is most useful when your question is one of these:

  • why is this source family in the repository at all
  • why has one source family been refreshed more aggressively than another
  • why did a public output change after a source update
  • why does the repository keep some material visible as context but not as strong evidence