About an annotated bibliography

Today an updated version of my bibliography of papers about distributed approximation is out (see here on the arxiv).

What it is

The idea is to have a list of papers, with a few items for each, telling what is the setting, the main result, and how it compares with the other papers. The origin of this one is a project for which it was important to understand previous work, and the literature was very confusing. So I did the first version of this bibliography for my colleagues and myself, and then when we finally abandoned the project I thought it would be nice to make it readable and public.

Note that it is not a survey: I am not an expert of the area, I don’t explain any proof technique, do not highlight any open problem. It is just an annotated list of papers.

Why it’s useful

I think that such documents are useful, and that we should spend more time writing them, even if they are not counted as publications.

First, it provides a good tool when starting a project in an area that one does not know well. It provides an exhaustive list of the papers of the area, which allows to understand (1) the history of a problem, (2) the state of the art (at the date of the latest update), and possibly (3) which papers are to be credited for a given technique. In this sense, it makes the area easier to reach, and avoids having only a small set of authors being able to understand what’s going on.

Second, such bibliographic documents allow for a more healthy citation culture in research papers. Without a bibliography to cite, one is often in a problematic situation for citing papers:

• Either you cite all the papers of the area that you know, and then you do it in a way that is useless for readers (long lists of citations, without any comments, and without the promise that it is meant to be exhaustive).

• Or you cite specific papers, and then maybe some people will be pissed of, but more importantly you might not cite the most relevant ones, but only the ones you see cited often. (I can see this clearly in my citation counts, I have some papers that are always cited as “there is also this other stuff done in area”, while others would actually be much more relevant. I guess that, when in a hurry, one just copies the citations from previous papers.)

• Or you cite basically nothing (which is more graph theory style) but then you do not give any help to the reader to understand the area, except for your result (and this 1964 paper of Erdös, in which you found the conjecture).