I mean, it is good they provide very cheap, very fast publication, far exceeding paper journals. It is just stupid that they do not also far exceed paper journals in the refereeing (=commenting) dimension. Which they easily could in terms of speed & volume, thus obsoleting paper journals. The alleged reason (they told me) for their refusal to implement commenting was it would have been "impolite". This presumably explains the arxiv overlay sites Mike Stay mentioned. Aha. Evidently created by people who do not care about politeness, thank god. But still stupid. Another aspect is this. A lot of papers do big computations, produce a lot of data, use some computer program, etc. This data or program could be posted. IN paper journals, it is not published since long. Electronic publication need not suffer that defect. But -- it is a fact, that in all, or perhaps 99% (I do not know), of ArXiv papers where they could publish such stuff as an auxiliary file, they do not. I do not know the reason, but it is a fact. And in some cases, this nonpublication makes their actually published paper be silly. I mean, it works something like this: OFFICIAL PAPER: we did a huge computation to find a few numbers. We'll offer some vague descriptions, but basically, just trust us, we got the numbers right since we're pros, so we have no need to provide any code or details. UNPUBLISHED COMPUTER PROGRAM: has full details. May have bugs, but since secret, you'll never find them. Historically, many papers on this topic, including sometimes by the same authors, have been erroneous due to bugs. IF YOU ASK THE AUTHORS FOR THE DETAILS: they won't respond. it is kind of science by reputation, not science by science.