Two different numbers, and only one of them is deletions. Flagged means a comment stopped appearing in the listing , all this tracker can see by itself, and it over-reports badly: a reply under a removed top-level comment gets hidden, not removed, and still answers the API when asked for by id. Confirmed means it was asked for by id and came back empty. Count deletions off the confirmed number; the flagged one is a starting point, not a finding.
Reading the chart below
It counts occasions , one account, or one wiped thread, losing comments in one scan , not comments. That total is not meant to match the confirmed count above, nor the separate actions figure, which counts something narrower again.
How long comments lasted cannot be answered on this video. Tracking started after the video was published, so only 2 of the 10 confirmed removals are comments watched from the moment they were posted. For the other 8 the age at removal is a floor rather than a measurement, and a distribution drawn from what is left would describe when this tracker was switched on rather than anything YouTube did. The chart is withheld for that reason.
Read the across axis carefully. It is when this tracker noticed, not when it happened. Scanning runs on a timer, so everything found in one pass carries the same timestamp. These 10 removals were spread over 8 distinct moments, and the busiest single pass accounts for 3 of them. Removals closer together than one scan cannot be told apart, and a vertical line of dots is scan batching , not evidence that anything happened at the same instant. The shaded band is everything posted before tracking began: a dot in there was already old when this tracker first saw it, so its height is real but its distance from the diagonal means nothing.
The three shapes, and the test behind each
These are groupings this page applies to the data. Each is a rule about what the API returned , none of them names a cause.
- Whole thread went , 0 comments in 0 occasions. The rule: the thread stopped being returned, and every reply in it that was checked individually also came back empty.
- One account, across more than one thread , 4 comments in 1 occasions. The rule: every comment this tracker ever saw from that author came back empty, and those comments sat in more than one thread. It is a rule about the set, not about timing , an account whose comments went on separate occasions draws more than one dot.
- No shared pattern , 6 comments in 3 occasions. Everything neither rule matched. 0% of it (0 comments) is accounts this tracker has only ever seen once, so there is no set for the second rule to test. The rest belongs to 3 accounts that do have a set, which did not span two threads.
Precedence: 0 comments meet both of the first two rules and are counted under the account shape. 1 accounts and 0 threads meet the rules.
Other measurements, and what each one counts
| Measurement | What it counts | Value |
|---|---|---|
| Comments by Professor Dave Explains, confirmed gone | posted by the channel that owns the video, asked for by id, came back empty | 0 |
| of those, inside a wiped thread | the first rule above also matched the thread they sat in | 0 |
| Comments by Professor Dave Explains, flagged but still answering | stopped appearing in the listing, still returned when asked for by id | 0 |
| Author still visible , wiped threads | removed replies whose author is tracked elsewhere on this video, share of those with that author still visible | 0% |
| Author still visible , mixed threads | same measure, in threads where some checked replies came back and some did not | 0% |
| Largest wiped thread | replies in it, distinct accounts among them, and how many of those accounts still comment elsewhere on this video | 0 / 0 / 0 |
Counts, not conclusions. What they add up to is left to the reader , read the limits below before drawing one.
What cannot be known from this data
- Who. Self-deletion, channel-owner removal and YouTube policy removal give the same empty response. The API never names an actor.
- Whether an account-shaped disappearance is the account or the channel. An account being removed and the channel owner hiding that user both take every comment at once and answer the API identically.
- Anything about a removal from the survival of its replies. Replies surviving means only that nothing acted on the replies. An owner removal that does not cascade looks the same as a self-deletion, and for the thread shape survival is how the group was defined , so reading a cause off it is circular.
- Whether it is permanent. A comment held for review answers the same way and can come back later. "Gone from the API" is what is measured; "deleted" is an inference.
- The exact moment. See the note under the chart.
- Anything that happened before this tracker started. It began watching part way through the video's life, so 8 of the 10 confirmed removals are comments that were already there when it arrived. For those, "how long it lasted" has no answer , only a floor , and any comment removed before the tracker started was never seen at all. Removal rates for the earliest posting days are therefore floors too, and the low rates there are as easily an artefact of arriving late as a real difference.
| Day spotted | Whole thread | One account | No pattern | Total |
|---|---|---|---|---|
| 2026-10-04 | 0 | 4 | 6 | 10 |
Where the gap comes from. The tracker marks a comment deleted when it stops showing up in the listing, and never confirms it. This job re-checks every flag against the API individually and caches the answer, so a confirmed removal is only paid for once. Latest run, 1 API calls: threads, nothing flagged. Individual removals, all 10 confirmed gone. Replies under a removed thread, nothing flagged. The flags are matching the API this run, so the two totals are close.
| Bucket | Rule the code applies | Comments | How sure |
|---|---|---|---|
| Top level comments removed | thread stopped being returned by the API | 0 | solid |
| Replies under those comments | reply in a removed thread, same timestamp | 0 | solid |
| Account wipes and bans | same account, 3+ removals, one minute, 2+ threads | 4 | likely |
| Spam filter, copy paste | identical text removed 2 or more times | 0 | likely |
| Creator or channel mods | removals in creator threads above the base rate | 0 | estimate |
| Everything else | no rule above matched | 6 | estimate |
What a separate action means. Comments removed from one account within the same minute are counted once, however many there were. So 5 is how many times something acted, not how many comments it cost. The bottom two rows of the table are estimates rather than counts: the API does not report who removed a comment, so they are derived from account and content patterns. The creator row comes from removals appearing in threads Professor Dave Explains posted in more often than the base rate, and is zero when there is no such gap.
Destiny Loves GenocideProfessor Dave Explains972 creator comments12,887 comments gone (7,542 confirmed)15,062 threads trackedupdated 04 Oct 2026, 20:39
Everything Is Crumbling For Destiny (And Pangburn) (deleted)Professor Dave Explains203 creator comments5,213 comments gone (5,213 confirmed)7,480 threads trackedno longer on YouTube, last seen 12 Sep 2026, 01:14
Everything Is Crumbling For Destiny (And Pangburn)Professor Dave Explains130 creator comments1,334 comments gone (881 confirmed)4,086 threads trackedupdated 04 Oct 2026, 20:49
This is What Bad Faith Debate Looks Like (Destiny Israel Fact Check)Professor Dave Explains19 creator comments10 comments gone (10 confirmed)2,022 threads trackedupdated 04 Oct 2026, 20:52you are here
The API stores replies flat: every reply’s parent is the top-level comment, with no record of which reply it answered. The only clue is the @handle prefix YouTube used to insert on a reply-to-a-reply, and it is almost never there any more. On this video 19 of 19 creator replies carry no prefix. Without it there is nothing to match against, so those replies sit against the top-level comment rather than guessing at a target. Open a collapsed run to read what was actually being answered.