Of 322 authors, the most prolific contributed 1% of the cases

By Alpha Lay · 2026-09-26

A common instinct when looking for partners or accounts to follow is "find the top accounts." We tallied the author field across the case library and found that this pool barely has a top.

Where the data comes from

The author field records the publisher's handle on the original platform. Of 1,274 records, 637 (50.0%) have a value — the other half do not, mainly because they come from GitHub repos, tutorial pages or aggregator sites where no single account can be attributed.

So everything below covers half the sample, and specifically the half that can be traced to a single publisher on a social platform. That is the most important caveat here.

Finding 1: 322 authors, and the leader has 1%

637 cases spread across 322 authors. The top four have 13 each — 2.0% of the attributable sample, or 1.0% of the full 1,274.

The top twelve total 123 cases: 19.3% of the attributable sample, 9.7% of the whole library. Put differently: take away the dozen most active accounts and you have taken away under a tenth of the cases.

Finding 2: Half the records have no author at all

This deserves its own section, because it determines how much the long tail above can be trusted:

Exactly half have no attribution. The reason is straightforward: a case may come from an awesome-* repo on GitHub, a tutorial's index page, or a now-dead aggregator — none of which has an "author" concept.

So we face a possibility we cannot rule out: the genuinely most prolific person may be sitting inside that 50%. If an account's cases were picked up by two mirror sites and three repos, each time dropping the author field, they would show up here as not existing.

Finding 3: A "many try, few persist" distribution

Sorting the 322 authors by case count:

About 87% of authors contributed only 1 to 3 cases.

This is a classic sampler's distribution — most people tried once or twice, a handful kept producing, but nobody produced enough to define the style of the field.

For a body of content like "AI video cases," that shape is itself meaningful: it says no authoritative source has formed yet. Anyone is replaceable, and nobody is essential to follow.

What you can do with this

If you are doing outreach or partnerships: by the data, following any single account yields very little (2% at most). Watching topical streams on the platform beats watching people.

If you are collecting cases: do not use "author" as a dedup or merge key. Half the records lack it, so using it will drop or wrongly merge half your data (which is also why we deduplicate on the original publication link). It also means: multiple cases from one person are likely logged as different sources, making author-level statistics inherently incomplete.

If you are a creator yourself: the flip side is good news — 13 cases puts you in the top four of this pool. Public supply is thin enough that producing a dozen or so entries makes you a repeat citation in indexing projects.

Method and limitations

• The sample is only 637 records (50.0%): conclusions apply to "cases attributable to a single account" and do not generalise to the library.
• Handles are not manually reconciled: different accounts may belong to one person (the similarly named ManuAGI 🤖 and ManuAGI01 very likely do). We did not merge manually, so "322 authors" is inflated.
• Display names are unstable: handles contain nicknames, emoji and Japanese descriptions (e.g. mayv@簡単プロ級プロンプト公開中！), so naive string deduplication understates one person's output. The two errors point in opposite directions and we cannot net them out.
• No follower or engagement data was collected: this counts cases only, which is not influence. Someone with 2 cases that everyone cites may matter more than someone with 13.
• All figures use the 1,274 index records.

Reproduce it

Case detail views include the author field where the original source provides one: https://cases.aishifu.shop/