This was the missing hidden variable not explained in the narrative.
Firefox was previously significantly behind Chrome in the technical design, now it's getting better but there is still some lag.
And as explained in the article, this lag has been historically to a large degree caused by the extension model (need to keep backwards compatibility on a huge API surface, impossibility to introduce multi process architecture, need to always negotiate with extension devs on what can be changed and what not...).
There were two options:
1) specialize on power users with XUL extension but lag even more behind Chrome - leads quite certainly to drop of users, slide to irrelevancy and the eventual death
2) try to catch up with Chrome in technical design, but drop historical baggage, including XUL based addons - does not guarantee success but still provides at least hope
> ... and stopped installing it on computers of friends and coworkers, so there must have been some small effect.
It's sad to see people supporting monopoly this way ...