YouTube only searches the comments already rendered on screen, which is why a term buried five pages deep never appears. ClearComments loads the whole set into your browser first — every top-level comment and every reply — and then searches locally as you type. Filter by author, by date range, by minimum word count, or narrow to the comments carrying a moderation signal.
The statistics view is built from the comments already loaded, so it costs no extra API quota: comment and reply volume over time at hour, day, week or month granularity, unique authors and comments per author, total and average likes, the most frequent words, the most active commenters, the questions your audience asked, and a flagged count broken down by spam, scam, insult, criticism and channel impersonation.
The scam that costs viewers money is a reply wearing the creator's name and photo. A display name proves nothing, so every reply is compared against the channel's real ID, and look-alike substitutions such as a zero for an "o" are folded before comparison. The genuine owner is marked; the copy is flagged.
Timestamps mentioned in the comments become a chapter list in video order, with how many people cited each moment, ready to paste into a description. Giveaway draws are seeded and reproducible, so the seed and the entrant list decide the winners between them and the result can be checked rather than trusted. Anything on screen exports to CSV, a real Excel workbook, or JSON with reply threading intact.
Public comments on public videos only — the same ones any visitor can read. Comments YouTube has already held for review are visible to the channel owner alone through an authorised login, so no public tool can show them.