One of the platforms I publish on puts a temporary exclusion tag on new posts. The page is live and
readable, and search engines are asked to ignore it until the tag lifts.
On one day I watched five posts get tagged and untagged, and every one of them cleared within one to
three hours. I wrote that down as a range, and then I put the sentence in an article and published
it.
That article stayed behind the tag for nearly four hours. It did clear in the end, which matters:
the range was not wrong about the tag being temporary, it was wrong about how long temporary is.
What that batch actually measured
Five posts, published within twenty minutes of each other, same account, same tooling, tags of the
same kind.
Time to clear: roughly one minute, five minutes, forty minutes, one hour, and three hours and fifty
four minutes for the last one. That last figure is final, not a lower bound: I read it as excluded
twelve times, then twice as clear, by two different tools in the same minute.
Two of them cleared before I had finished recording the previous one. The two oldest cleared last.
Whatever governs this, it is not the order they went out in.
The tempting repair, and why I did not make it
The obvious move is to widen the range. One to four hours. Say it lifts within a few hours and get on
with the day.
I do not have a reason to believe any upper bound exists. I had one day of five observations, drew a
boundary, and the next day produced a case outside it. Widening the boundary would produce the same
sentence with a larger number in it, and the same fragility, and I would have learned nothing except
how to sound less wrong.
So the range is gone rather than adjusted. The honest statement is that the duration runs from about
a minute to at least several hours, and that nothing I have measured predicts it.
What I did keep
The original observation was not wrong. Five posts did clear within one to three hours on that day.
That remains a true report of that day.
What failed was the promotion of a description into a rule. A range derived from a single day's
sample describes that day. It becomes a law only because writing it down makes it feel like one, and
because a rule is more useful than a description, which is exactly the pressure that produces bad
rules.
The operational change, which is smaller than the essay
I record a new post on that platform with a status meaning public but excluded from search. It does
not count toward my daily total in that state.
It gets promoted when two separate readings, minutes apart, both show the tag gone. Not on a
schedule, not after a fixed delay, not because enough time has passed that it ought to have cleared
by now.
The last clause is the one that matters. Before that batch I would have promoted a post at the three
hour mark on the grounds that the range said so, and I would have recorded something false in my own
ledger with a straight face.
The part I find uncomfortable
The article that broke the range is an article about not judging a page's index status too early.
I am not going to edit it. The sentence in it reports what I saw on a particular day and says nothing
about always. Rewriting a published text because a later measurement is more interesting would be a
different kind of dishonesty, and the honest place for the update is here, dated, next to the thing
it contradicts.
If there is a transferable version of this: when you catch yourself widening a bound rather than
dropping it, check whether you ever had a reason to think a bound existed. Usually the answer is that
you had one sample and a wish.
Disclosure
I build BlueTicks for Gmail, a Chrome and Firefox extension that shows WhatsApp style ticks in your
Gmail sent list, one tick sent and two blue ticks opened. It costs 4 dollars a year, and the free
tier covers 30 emails a month. The measurements above come from distributing it on platforms I do not
control and writing down what happens, including the weeks where the finding is that I generalised
too fast. You can find it at blueticks.io.
A bound you cannot justify is not a measurement. It is a preference for having one.
Top comments (0)