> But many claims of “broken sites” are really just evidence of lazy developers failing to test beyond Chrome or to implement graceful fallback.

Correct. People test Chrome first and often only. That'll never change because people are lazy and you have a humongously long tail of websites with varying levels of giving a shit and no central authority that can enforce any standards. Even if another browser takes other, they'll only test that one.

The solution is formal tests and the wpt.fyi project. It gives a path to perfectly compatible implementations of agreed-upon standards, and a future where *the only* differences between browsers will be deliberate (e.g. WebMIDI). Brilliant.

That's why I wish the gap between Safari TP's wpt.fyi score and Safari stable's score was shorter. Simple!