Telling a real difference from an apparent one

The procedure

1. Were both pages fetched just now? If not, fetch them and start again. This removes more cases than the other three put together.
2. Am I in the same signed in state on both? If not, fix that and start again. A signed out view is a different view, not a different market.
3. Am I looking at the same page, reached the same way? If not, get to the same place on both and start again.
4. Is the difference in substance, or in speed and appearance? Speed and appearance carry nothing. Substance, having survived the first three, is worth acting on.

Why the order is the whole method

Each question is cheaper than the one after it and eliminates more cases. Working in this order means a reader spends seconds on the common causes and keeps their attention for the rare one.

Working in the opposite order, which is what people do naturally, means spending attention first and then discovering it was a cache.

What each answer rules out

Fetched just now

Removes stale pages, left open tabs and most differences in figures.

Same signed in state

Removes empty inboxes, missing content and a large share of apparent removals.

Same page, same route to it

Removes the adjacent page problem, which is common on a route you use rarely.

Substance rather than style

Removes speed, layout and everything else that is about the journey rather than the destination.

The one case that survives

A specific, identifiable difference in substance, on two freshly loaded pages, in the same signed in state, reached the same way. That is the case handled in the difference that matters, and the response there is to check the string rather than the page.

What this procedure is not for

It is not for deciding which route is better, because none of its answers bear on that. It is not for deciding whether the market is healthy, which no comparison between two of its routes could establish.

It answers one question: is this difference an artefact of my own looking, or not.

Why the procedure is short

Because a long one would not be used. The situations in which a reader needs this are situations where they are already mildly worried and want an answer in a minute.

Four questions, in a fixed order, is about the length that gets applied in practice. Anything more elaborate is a document people read once and never use.

The thing to take away

Almost every difference between two routes is a fact about your own looking. That statement sounds dismissive and is not. It is what makes the rare exception legible, and the exception is the only part of this section that is worth acting on.

Where this leaves the reader

With a comparison tool that mostly returns nothing, which is the correct behaviour for a comparison tool. A reader who expects differences to mean something will find meaning constantly. A reader who expects them to be artefacts will be right nearly every time and will still catch the case that matters, because that case is the one that survives four cheap questions.

What to do when the procedure comes back clean

If all four questions are answered and the difference has gone, nothing happened and there is nothing to record. That is the outcome nearly every time and it should feel like a non event rather than a relief.

If it survives, you have something worth acting on and a clear next step. Either way the reader ends up somewhere definite, which is the main thing a short procedure buys over an argument: it terminates.

What this study is not. It is not a diagnostic for anything beyond apparent disagreement between two routes, and it says nothing about why a route might be quiet.