About companion papers for reviewers

It is not a new observation that TCS papers are mainly written for reviewers. This makes sense, given that researchers want to optimize the list of accepted papers. Of course it makes little sense scientifically, since most readers will not be reviewers (hopefully!).

As a reviewer, I like that the authors explicitely discuss the novelty and difficulty; but as a normal reader, I just want to understand the content and get the big picture. A typical example is a paper that uses notions that are close to another area of research. As a reviewer I want to be sure that the authors are aware of it, and have checked the literature, but as a reader, I will just trust the authors (and the reviewers).

In some sense the discussion of the novelty and difficulty should be managed in the same way as the boring proofs in journal papers: you want them to exist, you want them to be checked by reviewers, but that’s not what you want to read.

Unfortunately many TCS papers appear only as 12-page conference paper. Sometimes I read a celebrated paper, which is painfully explaning why what they do is great, (of course it is!). It’s annoying.

We could imagine companion papers, where one would write specifically to the reviewers to explain in detail the novelty (with an in-depth review of the literature), and the technical difficulties (with a discussion of how all the known tools fail).

I tried to have such things in the appendix of a paper, but even though there were several warnings in the text, a reviewer did not read it, and said that the technique was well-known (when it was not). Maybe having it as a seperate text would avoid this.