Andrew Ollett’s justification of diacritics in dissertations is exactly right.  They are not that hard for nonspecialists to handle, but provide sufficient information for them to keep track of key terms and names.  It is hard to imagine a dissertation dealing closely with Sanskrit-medium philosophy while dispensing with the tools for representing the language unambiguously.  Jeff, send that dissertation back and have the student make it presentable.

Dominik, I think there are very important reasons for transliterating in discussions of Sanskrit and other Indian texts.  Apart from the dizzying variety of scripts involved in the primary sources, transliteration is a very useful aid for students and other non-specialists.  I recall that even before I learned Sanskrit, I built a repertoire of vocabulary from coursework and reading of secondary sources that allowed my first steps.  Using only Devanagari would please many north Indians, but it would shut out any non-Sanskritist readers in other countries.  And showing word breaks is another useful service.  (A stronger case can of course be made for printing full editions in Devanagari.)

Anyway, it is just not true that secondary-source publications on citing words in Greek, Russian, Armenian, Hebrew, etc. never transliterate.  It just depend on how narrow an audience is envisioned.

Best,

Tim