Why Unknown Devices Deserve the Full Experience

Detection systems commonly serve a simplified page to clients they cannot identify. The reasoning behind that default is sound and the outcome is the opposite of what it intends.

Unknown skews new, not old

A detection dataset describes what existed when it was assembled. Clients it fails to recognise are usually ones that appeared afterwards.

Old browsers, by contrast, are extremely well described. They have been in every dataset for years and match reliably.

So the unknown bucket fills with recent releases, unusual new devices and clients built on current engines, not with obsolete software.

The conservative default inverts the intent

Serving a minimal page to unknowns is meant to protect users of limited clients. It instead delivers the reduced experience to users on the most capable software.

Those users have no way to signal that the assumption is wrong. They see a degraded page and generally conclude the site is poor.

The error is invisible from the server, because the reduced page is served successfully and nothing is logged as a failure.

Degradation should require positive evidence

The safer rule is to serve the full experience unless there is a specific signal of a limitation. Absence of recognition is not such a signal.

Positive signals do exist: a failed capability test, an explicit user preference for reduced data, or a known defect in an identified version.

Each of those is evidence about the client. Failure to match a pattern is evidence about the dataset.

The full experience must survive being wrong

This default only works if the full page degrades gracefully when something is missing. That is a property of how the page is built, not of detection.

Building on a working baseline and layering enhancements behind capability tests produces that property naturally. Each enhancement fails independently.

Pages built the other way, where the full experience assumes every feature is present, cannot safely adopt this default and need detection to protect them.

The cost of each error direction

Serving too much to a limited client produces a page that may be slow or partly non-functional, and the user can usually still complete the task.

Serving too little to a capable client produces a page that works and is worse, permanently, with no path to the better version.

Comparing those outcomes is the whole argument. The recoverable failure is preferable to the silent one.