1. Usually, reference letters compare scholars at other institutions at roughly the same stage in their careers.

2. On fields in which a book is the standard contribution, usually impact and contribution is not well indicated by citations, for promotion purposes. It is indicated by book reviews, how many and their venues. Main journals in the field vs. specialized ones, many vs. one or two. In the journal article fields, citations indicate something about impact, but for substance, it is useful to rely on the letters.

3. Books tend to take a while to be influential through citations in journal articles, even longer in other books, times too long compared to when decisions are called for. Often, one has to rely on reference letters for first books when promotion to associate is on the agenda.

4. For journal fields: Citations are helpful for promotion to associate, crucial for promotion to full. Again, issues of delay, but much shorter delay than for books.

5. Some fields demand a book and several articles. (Usually, this is honored in the breach, by the way.) The articles here are previews, almost always. People are waiting for the book to really evaluate the work.

6. In a field like political science or sociology, while there is often talk about demanding a book and a bunch of articles, individuals tend to show one tendency OR the other. The articles can be incremental contributions, to be evaluated as a whole, or previews of the book. If previews, we want to see the book.

7. In all fields, I am convinced that it should be possible to explain to a professor layman and the provost what the contribution is and its significance, even if the field is technical or has a specialized diction and discourse. Moreover, letters can be substantive using lay language in describing the contribution clearly.

For a university to become stronger, MANY of the lateral appointments must be very stong, as should be MORE and MORE of the internal promotions. They set the standard, they benchmark. Strong ones cases make weak ones seem untenable, unless there are institutional reasons that are overriding.