Experts we follow

The weekly digest (curation.md) watches a set of people who write about this KB's topics. This page explains how that list is chosen, so anyone can check it, re-run it or challenge it. The data is in experts.yaml, and the tables below are generated from it. The method is the same in every KB of the family; each KB scores people against its own topic.

Last scored 2026-10-18: 1 tracked (1 vendor-staff), 1 on the watch list, 0 declined, 0 nominated and waiting to be scored.

This KB's study

First run 2026-10-18. Few people outside Liquidware publish on Stratusphere, so the list is short: searches of the web, Liquidware's blog and support center, GitHub and the sibling KBs' expert lists turned up no independent author with recent, original Stratusphere content. The vendor's own feed is tracked as the control source, and its GitHub organization is on the watch list. Nominations are welcome through the Suggest an expert to follow issue form.

The method

It works like a small systematic review: the rules are fixed before anyone is scored, they're applied the same way to everyone, and every score points at evidence.

1. Find candidates in several places

One list brings its author's blind spots with it, so candidates come from several channels, and each entry records which ones found it (found_via):

• The vendors' own channels: docs, sample and course authors, and people they feature
• Speakers at conferences and meetups on the topic
• GitHub: maintainers of tools people use, and the KB's prior-art searches
• Blogs and newsletters that post on these topics more than once
• Snowballing: who the people already on the list link to, cite or present with
• The sibling KBs' expert lists (family.yaml); use the same id for the same person
• Nominations through the Suggest an expert to follow issue form

2. Apply the gates

A candidate must pass all four gates to be tracked. Organisations go through the same gates and scores as people. Failing identity or originality declines them; failing only recency or feed puts them on the watch list if their score is high enough.

3. Score five things from 0 to 3

The total is out of 15.

• Tracked: all gates pass and the total is 10 or more. They get a [[source]] in sources.toml with expert = "<id>".
• Watch: identity and originality pass and the total is 6 or more. Re-checked each review, not in the digest.
• Declined: everything else, with the reason, so the same person isn't re-researched for nothing.

scripts/kb/validate.py enforces these rules: an entry's status must match its gates and scores, every tracked expert must have a source, and every source with an expert must be tracked.

Affiliation is recorded but not scored. Vendor staff often know the product best, so they aren't marked down. The summary above shows the mix, so a list that's all one company's voices is visible.

4. Measure and review

• Yield: the monthly freshness check reads recent digest issues and counts, per tracked expert, how many of their items were ticked as worth adding. A source that lists nothing, or only items nobody ticks, is flagged to re-score or narrow with keywords.
• Cadence: everyone is re-scored every 90 days. The freshness check lists anyone not re-scored in 90 days and any nomination waiting to be scored.
• Nominations: the issue form adds the person as nominated. When a maintainer labels it kb-accepted, the feedback workflow opens a PR, and the maintainer scores them in that PR.

What's recorded about people

Only what they publish professionally: the name they use, their public links, and their affiliation if they state it. No emails, locations or other personal details. Anyone listed can ask to be removed by opening an issue.

Tracked

Watch list

Declined

None.

Nominated, not scored yet

None waiting.