There will soon be a HTML native method to retrieve a device appropriate
<img>. It too will repeat the very same environment tests that both our CSS and JS have performed in order to modify their behaviour to the client’s capabilities. And in the case of HTML it will repeat these environment tests for every single
<picture>on the page.
This is incredibly inefficient, and has the potential to make hand-authoring web documents implausible. Not to mention the problem that’s really to blame: Our technologies need to sense the environment in unison and adapt to a design, not to a device.