It is fundamentally designed around a bad idea.
That seems rather extreme. What specific bad idea do you mean?
(Context here suggests that it's something like "the idea that typographical choices for LW2 should match those for the web as a whole", but even if LW2's design makes that assumption and even if it's a bad assumption it doesn't seem fundamental enough to justify your last paragraph.)
Context here suggests that it's something like "the idea that typographical choices for LW2 should match those for the web as a whole"
The idea that the study of typographical choices for the web is a mature science whose (nontrivial) recommendations can all be taken at face value.
What are your first impressions of the public beta?