Syllabic notation is harder to understand, someone talked about it above, syntactical boundaries are harder to parse, and the language’s functional parts - things like tenses and inflections, are mostly written in kana, the rest mostly in kanji.

This article (featured previously on HN) discusses the efficiency thing a bit - https://yakkomajuri.com/blog/language-efficiency - though I think the pixel count metric is meaningless and leads to weird conclusions, there are some good thoughts.