ECCC (both the Trier and HPI mirror sites) seems to have been down for the last day or so. I checked downforeveryoneorjustme, and it's not just me. I have some evidence (in the form of this twitter update) that it was up within the last couple of days, so I'm not especially worried about its long-term status. (In fact, that update was why I bothered to check out ECCC in the first place — the paper title intrigued me.)
I am worried, however, by the fact that it's been down so long without any indication that I can find for why it's down or when it's expected back up, and also by the fact that (up to this post) nobody in the theory blogosphere seems to have noticed or cared. If arXiv went down for a similar time period without notice I think there would be a lot more screaming. Is this evidence that people have been moving away from ECCC to other venues for online publication such as arXiv? Are the complexity theorists who tend to frequent ECCC the type of person who suffers in silence when something like this happens? Or maybe (as a note in this blog post from a year ago hints) it's often down, so by now the fact that it's down again is not news?
Or maybe it's just because of the weekend thing...
Maybe the problem indeed was not global.
Which URL are you using to connect to it?
I'm using http://eccc.hpi-web.de, but indeed, now the server is down! This is certainly a problem, since the alternative, which is ArXiv, is worse, as it doesn't have any filtering mechanism.
That's not completely accurate. ArXiv does have filtering mechanisms: it filters out papers that are off-topic or that aren't written like scientific papers (no bibliography, for instance). It doesn't filter out the obviously-bogus but on-topic papers, but those are easy enough to ignore and there are not so many of them as to overwhelm the others.
So what arXiv's filtering mechanism really seems to be lacking compared to ECCC is not either of those aspects, but rather the ability to filter the interesting papers from the boring ones. But are complexity theorists really so homogeneous in their interests that ECCC can do that sort of filtering accurately?
I disagree. All P vs. NP crank papers in ArXiv are "not written like a scientific paper", yet they are not filtered out. Another thing is that about half of the papers appearing in the computational complexity part of ArXiv are not really on topic (that is, they are not on computational complexity per se, but only marginally). These factors make ArXiv a place which is by far less useful than ECCC. Also, to sort out the papers not on-topic is not an easy task on first glance, thus you loose the cognitive-psychological factor of having a list of cutting-edge on topic papers in computational complexity, like in ECCC.
Site outages are not the only thing about ECCC that users have suffered mostly in silence. Last I checked (in May), we:
- can't preview the use of Latex in abstracts, so many fail to compile properly;
- can't edit abstracts & keywords once they're up without posting a paper revision;
- can't get pdf files to have properly functioning hyperref links.
I've complained to the administrators about these things and they are aware of them, but the problems seem to persist. They are nice people and will provide help on an individual submission if you request it.