Hijinks and outrage on Twitter today, after the discovery of a spider trap embedded in the pages of the ACS journal platform.

It first came to light via Peter Murray-Rust's blog, where he reported that the enigmatic ‘Pandora’ had found herself and her institution blocked from pubs.acs.org in its entirety, after first uncovering and then, understandably curious, clicking on the poisoned link.

The result? This message:

Image from PMR's blog

This link in question looks like an article DOI, presumably so that a web crawler attempting to aggregate ACS articles would gobble it up, before finding itself unceremoniously banned.

It's not visible on the journal pages themselves; rather it's hidden in the underlying HTML. However, once found, a simple web link can get anywhere.

Judging by the Twitter reaction, a number of people followed the link from the blog post, either not realising the consequences, or seeing it as online activism to highlight the ACS's behaviour. Cue blocked institutions, with Nature Publishing Group among the victims.

Things kicked off further after Ross Mounce, perhaps somewhat ill-advisedly, included the link in a tweet without much context. Angry reactions followed, as more people clicking the dread link found themselves unwitting activists, and found their institution's IP range out in the cold. That said, a large part of that anger was (rightly in my opinion) directed towards the ACS.

Alarmingly/amusingly, depending on your point of view, it was briefly feared that the entire Portuguese university system had lost access, due to its consortium subscription. Apparently not.

By this point a hashtag, #ACSgate,1 had emerged.

Also emerging was a degree of confusion as to what the link was and how it got there. I'm not going to post the link itself (don't want any more casualties), but here's what it looks like:

Image from PMR's blog

Although not, as pointed out by CrossRef, a true DOI, it's in the standard format of the ACS platform, e.g., http://pubs.acs.org/doi/abs/10.1021/es405402r (the level after ‘doi’ is ‘abs’ or ‘pdf’ or ‘full’ depending on whether it points to the abstract page, pdf, or HTML full text). What's odd, though, is that it has the DOI prefix 10.1046. As can be seen in the standard ACS url above, their prefix is 10.1021. 10.1046 belongs to Blackwell, now part of Wiley. And the landing page that the DOI points to apparently belongs to yet another publisher, Informa.

On Twitter, CrossRef surmised that this meant it was a prank. However, since link-clickers were not simply seeing a warning message, but were finding themselves comprehensively banned from pubs.acs.org, that seems unlikely. While these things are hardly my specialist subject, it seems that it would be difficult enough for an outside agent just to slip this link into HTML of the ACS platform, nevermind imposing a block on those who followed it.

Presumably after receiving record levels of correspondence to ipblock@acs.org, the ACS issued a response of sorts, in the comments of PMR's blog:

Thank you for alerting us to the finding shared by your reader. We are exploring and are committed to providing text and data mining solutions for readers of our open access content. In the meantime, for those who have unfortunately clicked on the link referenced and received the spider message, please email support@services.acs.org with your institution name and we will work to reinstate access at your institution as quickly as possible.
Darla Henderson, Ph.D.
Asst. Director, Open Access Programs
American Chemical Society

This touches on another issue: aside from the clumsy and irresponsible way that the ACS have blocked any attempt at data mining, the paper that originally triggered the avalanche was open access. Why are the ACS blocking data mining of a paper when the authors (or their funders) have paid to make it freely available?2

All in all, it seems like a massive internet fail on the part of the ACS. Despite lacking any particular expertise in such things, the failings of their spider trap are not difficult to see. I'm told that there are better and more responsible ways to achieve this (where appropriate! Don't try to lock up open access content), and I trust that the ACS are currently thinking hard about some of them.

(More from PMR here and here)

1. I hate the lazy way that appending ‘gate’ to any word denotes ‘some kind of scandal involving this thing’, but I grudgingly concede its utility in the condensed medium of Twitter.

2. It's not clear (at least, it's not clear to me) which Creative Commons licence it was published under, but since one of the funders is the Wellcome Trust, it ought to be CC-BY.