Measuring is not seeing

"Difference: 0 pixels." The measurement was correct, the conclusion was wrong. What the person saw on screen had nothing to do with it. A small lesson on the difference between measuring and seeing.

In short: a display correction announced with an impeccable number — 68 pixels of difference brought back to 0 — while the visitor, for their part, was still seeing the old display and missing headers. The number referred to the correct file. Simply, it wasn't the page they were looking at.

1. The measurement that reassures

Correcting an alignment without a browser is tempting: you open the file, count the elements, calculate the positions, announce a result. It's fast, it's reproducible, it produces a nice number. And a number looks like proof.

On the page listing articles, the measurement was telling the truth: the titles on the same line did indeed all start at the same height after the correction. Except the person looking was seeing something else — because they were looking at production, and the correction was awaiting approval in the trial.

2. The three things a measurement doesn't say

It doesn't say which page is being served. A corrected file on the disk, a copy published elsewhere, a cache in between: three possible states for a single address. The measurement applies to the one you opened, not the one you serve.

It doesn't say what broke nearby. You measure what you decided to measure. The title alignment was perfect; meanwhile, a header image had disappeared from some cards — outside the measurement's scope, therefore invisible to the report.

It doesn't say what the page actually becomes. A page waiting for a blocked external resource never triggers its load completion event; a language toggle script can replace the page before you look at it. Two pitfalls encountered on the same day, and two measurements that came back empty or referred to the wrong version.

3. The rule we adopt

As soon as a correction affects what is visible — alignment, header, image, layout, text — we open a browser on the address the person is really looking at, we look, and then we announce a result. In that order. The measurement retains its full usefulness: it tells by how much we moved. It never tells that it's good.

And when we announce, we specify what we opened: the trial or production. "It's fixed" and "it's fixed, live" are two different statements. Confusing the two is asking someone to check on your behalf — and to discover your mistake before you do.

4. Why this rule is written down, and not just understood

A rule that lives in someone's head gets lost at the first change of context. This one is recorded where it will be re-read before the next action, along with the reason that gave rise to it. It's the same principle as the rest of the tower: what isn't written where it will be re-read will be paid for again.

Measuring is counting. Seeing is opening. Both are necessary; only one of the two is enough to be confidently wrong.