A stale page is not a stale market

The belief, stated fairly

The belief is that the three addresses are updated separately, so one of them can lag. It is not a stupid belief. It is exactly how mirrors worked in the older sense of the word, where a mirror was a copy of a set of files that got refreshed on some schedule.

Every intuition a reader has about the word points this way, which is why the belief survives being contradicted.

Why a route cannot lag

A route does not hold anything. It is a way for a request to arrive. There is no store attached to it that could be refreshed early or late, and no schedule on which it would be.

If two routes reach one market, they reach it as it is at the moment the request arrives. The idea of one being behind the other requires a second copy of the market, and a second copy would make the two different markets rather than mirrors.

So where does the old page come from

From between you and the route, almost always from your own browser, which is the subject of your own cache. A page you have seen before can be shown to you again without anything being asked of anybody.

It can also come from time. Two pages loaded twenty minutes apart are two different moments, and anything that changes will differ between them for the least interesting reason available.

How to tell the difference in practice

1. Force a fresh load rather than returning to a tab you left open.
2. Do both routes within a short window so that time is not the variable.
3. Look at something that changes visibly, and check it in both directions rather than in one.
4. If freshness fixes it, it was never about the addresses.

The damage the belief does

A reader who thinks routes lag will start preferring the one that seems most current, which is a ranking built entirely out of their own browser's behaviour. That is one of the routes into assuming a ranking.

Worse, they will read the lag as evidence about maintenance, and from there conclude something about whether the market is being looked after. A chain of inference that starts in a cache should not end in a judgement about anybody.

The version of the belief that is true

One thing genuinely can be out of date across addresses, and it is not the market. It is the set itself. A published list of addresses can be older than the arrangement it describes, which is the whole of the lists section.

So a reader is right to think staleness is a real problem in this subject. They are usually attaching it to the wrong object.

What to do with an old looking page

Treat it as a browser event, not a market event. Reload properly. If it persists across a genuine reload on more than one route, the thing that changed is the market itself and it changed for everybody, which is not divergence at all.

Why this study exists at all

Because the wrong model is unusually durable here, and because it generates confident conclusions. A reader who believes in lag will explain almost any difference with it, will never test it, and will never be corrected, since nothing they can observe distinguishes lag from a cache. Naming the alternative is the only way out.

What the correct model predicts instead

If routes cannot lag, then differences in freshness are always about the reader's side, and they should be fixable by the reader. That is a testable prediction and it holds up: a genuine reload removes the difference nearly every time.

A model that makes predictions you can test in ten seconds is worth adopting on those grounds alone. The lag model makes no testable prediction at all, which is precisely why it survives so long in so many readers.

What this study is not. It is not a claim about how often anything on the market changes, which this site does not observe.