Her main points are basically unified theme and form follows function and how McMansions often break those rules (a hodgepodge of different elements, columns too large for the things they support).
Her blog also informs fundamentals of good taste in architecture. It's no different from learning what makes a classical painting good or good pacing in movies.
>... that's it. No explanation [...]
The explanation is ~'it shouldn't overwhelm it'. Of course that's a value judgement. Like saying the front bumper (aka "fender") shouldn't be too big on a car; or a persons hat shouldn't be too big for their hat (tell that to JK, https://goo.gl/images/ztdusQ).
The implicit point is that the transom is ostentatiously large, enlarged beyond the reviewer's sense of propriety.
Surely there is nothing else. It's a judgement of style?
The second is poking the US tradition of "I have money and therefore I'm right about everything, even when I have no idea what I'm doing" squarely in both eyes.
For the first, there is an embryonic science of aesthetics. Features like self-similarity, scale coherence, integration, and symmetry tend to score more highly than random content blobbery and mash-ups.
For the second - of course it's political. But in a country where freedom of speech is supposed to be a thing, I see that as a feature, not a bug.
Worse is that most of their advice is actually harmful. For instance they don't like windows, especially asymmetrical ones. But lots of scientific studies have found that windows improve mental health significantly in a number of ways. Sunlight is very good. Or they demand houses be symmetrical. But if I've learned anything from games like dwarf fortress and Minecraft, that's not true. While symmetry might look nice, it generally creates designs that are less practical than if you weren't constrained by it. And every rule they have is like that. Just purely aesthetic constraints that trade off against other more practical values.
I agree with practically all of its conclusions (ugly, oversized, out of place, generic, etc.), but the entire thing was a bunch of worthless value judgements: this is bad, that is bad. Not: this is bad because of function, but literally "this should never happen" -- no explanation.
Real example from way back machine, from the blog's most recent post there:
... that's it. No explanation, just "I'm right, they're wrong." It's just lousy, valueless criticism. Nobody would take this lazy opinion drivel from any other kind of critic.(Separate: Do I think zillow was right here? Of course not).