When Open Access Fails

I’m fairly busy as of late and so I regularly set aside some weekend reading, such as Sergey Gavrilets‘ new paper on investigating the impact of egalitarianism on human evolution during the Pleistocene. The paper was published the other day in the open access journal PLoS One. Yesterday, I bookmarked the paper’s DOI (10.1371/journal.pone.0003293), in hopes that I can read it today.

Fast forward to this afternoon… I sit down to download the PDF, print it out, and read it. I know that DOI’s have several layers of resolution and PLoS’ DOIs resolve to plos.org. I click my bookmark and much to my surprise am presented with this on my screen:

PLoS.org Expired

Ahh, the ever familiar landing page of shame, a billboard to the public pointing out those who forgot to renew their domain name with their registrar. For those that don’t know what I’m talking about, I’ll simplify it. In the land of the internet, there are special servers, called nameservers which handle the association of a domain name to a computer’s [IP] address. Without them, we’d have to memorize numbers like to access Google. To use these servers, those of us that own domains have to pay a regular annual fee. If we don’t, the service will put up this landing page. After some time, the name will released into the market and other’s can grab it.

I don’t know how long this problem will last for plos.org, but even if it lasts just for today or this weekend, it is an embarrassing problem for the Public Library of Science. It also raises my eyebrows and makes me question the responsibility of the individuals who run the joint. The Public Library of Science is completely web-run operation. If they can’t stay on top of renewing their domain names, how long can they be the premiere open access publication? We can’t visit their homepage, thousands of DOIs aren’t resolving today and I wonder if emails addressed to *@plos.org have been bouncing back?

There could be obvious explanations, though. Maybe the guy who’s in charge of managing the domains is vacationing? Or maybe they actually switched some settings, such as the IP address the domains point to. That could definately be the case. Come to think of it, since their other services that live on subdomains, such as PLoS Biology is working fine, I hope that’s why. But, I honestly couldn’t access the journal via DOI all day long. When IP addresses change, it doesn’t take that long to propagate across the whole world. So I decided to do a quick WHOIS search on plos.org, and in fact, the domain name did expire on October 2nd, 2008 and was not renewed.

Anyways, no hard feelings for the PLoS guys… It is hard running a nonprofit organization. I know, I used to work for one. In my experience, there never seemed to be enough people and everyone wore different hats. I really love their journals too. They’ve done some amazing things, such as the comments and trackbacks in PLoS One let alone the open access initiative. But I really would like them to show a little bit more professionalism, especially with handling their flagship domain, plos.org. Oh yeah, let me remind you that PLoS had performance problems with their content management system that handled all their journals earlier this year. The problem would slow their site to a crawl, driving readers away in frustration over wasted time. It is fixed now, but it took months!

P.S. I ended up finding that article the ‘hard way’ via clicking around plosone.org.

12 thoughts on “When Open Access Fails

  1. It was Saturday, when the online traffic is down 50% and only the most grizzled users are surfing the Web. The domain went down, the IT team got to it and fixed it within a couple of hours. What’s the problem?

  2. Uhh… are you seriously asking what the problem is? Bora, I know you’re the online community representative and try to save face for PLoS but again are you really asking that question?

    To answer, there shouldn’t have ever been a downtime. You know that should have been the case, and I am surprised you asked what the problem is. The problem is downtime and that is never good for any webpage. I can’t believe I have to tell the online community manager that downtime equals bad.

    The domain should have been renewed before the expiration date to avoid downtime. I know how registrars work. For starters they let you buy years at a time, so that you do not need to renew every year. Secondly, registrars send reminder emails to the registrant that their domain is about to expire well in advance of the expiration date. The fact that it did expire tells me that the plos.org ignored these emails. And that’s why I said plos.org dropped the responsibility ball.

    Also, plos.org wasn’t down just for a couple of hours. A couple of hours is 2-4 hours. I mentioned the domain expired on October the 2nd and I saw the domain was down all of October 3rd. That’s 24 hours.

    Again, to reiterate, even though the downtime occurred on a day when online traffic is regularly down, what do you have to say to the other 50% of visitors who did try to visit your page and saw the splash screen? I checked out the Alexa traffic rankings for plos.org and, on average 37,000 visitors hit up plos.org a day. If traffic dropped to half that — say 18,750 — that’s roughly 19,000 visitors that saw that.

    Is it okay for a webpage to have any downtime? No, it is not. Enterprise companies like Amazon get it. Uptime is their bread and butter. PLoS should get it too. Part of your organization’s business model come from publication fee, paid by the author. If I were an author who had a publication hosted on plos.org, I’d be fairly upset that they couldn’t serve up my paid-to-be-published paper cause someone decided to snooze on the renewing the domain.


  3. It happened to Google. Every technology occasionally fails. We were on top of this and fixed it as fast as possible. It was not a PLoS problem but a DNS problem. This has nothing to do with OA, as you could get to all the papers – pages that were down were the homepage and blog.

  4. Yeah, it happened to google.de, and Microsoft’s Hotmail. So does that mean since it happened to the big boys it’s OK that it happened to PLoS? If anything PLoS shoulda learned the lesson from others.

    And I disagree, this downtime does have something to do with open access. If the site is down, how open is the access if people can’t read the content? A similar analogy could be made to old fashion libraries. They are more or less open access institutions, but should the library not pay the electric bills and the computers that handle checkout system don’t work — how open is the access?

    The problem also was more deep than two parts of the plos.org online footprint. I mentioned in my second paragraph of the post, the pages that were down were not just the homepage and blog but every DOI link that referred to a PLoS paper, since the DOIs resolve thru plos.org. If anyone on that day clicked a PLoS DOI link they’d see that page. You know better than me how many papers the PLoS journals have published, but effectively thousands of DOIs did not resolve because the PLoS domain wasn’t renewed.

    One last thing, Bora. This was a PLoS problem, not a DNS problem. I’ll say it a million times, if the domain was renewed before the expiration date there would have been no downtime. Who’s responsible for paying the renewal fees of plos.org? PLoS is, of course!


  5. DNS was paid on time.

    Now I see FriendFeed has been down for a while… ;-)

    I am the total online maniac and I was not online. An alert reader told me that PLoS Blog could not be reached. I alerted the HQ but they were already dealing with it and the problem was much more serious than non-payment and it was not on our side. I opened PLoS ONE easily and opened a couple of papers easily. Ads/banners were off, the homepage of PLoS.org was off, the blog was off and some internal invisible pages were down. Only a tiny sliver of traffic to papers comes through the front page or blog – they most come through individual links. I have looked around and you were the only person trying to get a paper at the time, at least by using DOI-resolution as a method. And even you found the paper within minutes by switching to an alternative method – which is what redundancy in the system is for: it is designed to enable you to get to the papers even if one of the methods is temporarily unavailable.

  6. I noticed it was down as well, but was still able to grab a couple of papers without much trouble…

  7. Bora,

    You really know how to belittle the impact of downtime. Most organizations, such as Amazon, which faced a similar downtime this summer, issue responses like, “any amount of downtime is unacceptable,” and humbly apologize to their users and community. They issue a public and transparent plan of action to avoid such outages.

    I see no such response from you, just a lot of dodging the issue and passing the blame. This problem wasn’t isolated to just me, as you can see that Afarensis also noticed it as did Razib. We’re savvy enough to have found the papers we were looking for… but what about all those DOI links that resolve thru plos.org, i.e. such as the ones archived thru researchblogging.org? You really think that the impact was negligible?


Comments are closed.

A WordPress.com Website.

Up ↑

%d bloggers like this: