Fulfillment

I really enjoyed reading Why Sci-Hub Will Win. One of the key points is that the user experience associated with institutional access to journal articles can be pretty lousy. That’s not really news, but it is a nice reminder because I’ve been doing a lot of thinking about ebooks. With ebooks, the current goal seems to be pursuing an experience that isn’t demonstrably worse than that of using journal articles.

The Best Case

So in that Sci-Hub piece, Heathers spells out how much easier it can be to get something through Sci-Hub compared to institutional access. Of course, my initial knee-jerk reaction was to think of how the best-case scenarios for institutional access are better (and they are).

Heathers is talking about getting access to a known item, and his example is indeed laborious. Starting with a journal level search introduces a number of hassles. Many institutions offer two methods that will be a lot easier.

1. Google Scholar

Most academic libraries track their holdings/access entitlements/whatever you want to call them in a link resolver knowledge base. Typically, this can be pushed out to Google Scholar, and many libraries do this. If you’re on campus, you’ll automatically see links from GS to library full text. If you’re off campus, you would have to add those links in your GS settings. Basically you search for an article title, get a resolver menu, click the full text, and read the article. You don’t need to copy and paste a DOI.

2. Discovery Tool

A lot of academic libraries have a discovery tool, typically from Ex Libris/ProQuest or EBSCO. This would also work in most known-item searches. Like GS, it draws on a big store of data from publishers, so you can just put in an article title and get the result for your article. You may even get a direct full text link, skipping the resolver menu step.

The Many Opportunities for Failure

And wouldn’t it be great if these methods actually worked consistently? Let me spell out an incomplete list of ways these can fail.

1. Desired article is not in discovery tool (more likely if article is older).
2. Google Scholar passes wrong date to resolver, resulting in dead end (surprisingly common).
3. Either discovery option passes any number of incorrect pieces of information to resolver, resulting in dead end.
4. Discovery info and intended full text source have metadata from different versions (print vs. online), metadata differs, results in dead end.
5. Proxy server is missing config for full text site, or other proxy weirdness occurs.
6. Full text site is one of many that only links at journal or issue level rather than article level. User must repeat their search.
7. We have access to the journal through a database that doesn’t actually have the rights to reproduce everything due to author agreements, so we don’t actually have this particular article.
8. If the researcher goes through our discovery tool, they may very well be told that we don’t have access when an OA version is available.
9. Some perfect storm of proxy server + publisher platform + browser issue + PDF reader occurs.
10. Any one of the servers involved with this has a bad day.

And there are actually more ways for this to go bad. I know because at some point in the past 6 years, I’ve received reports of all of them. None of these problems is particularly rare either.

What to Do?

There are two key problems that lead to most of the problems above. One is that we’re dealing with many disparate systems. We have a huge network of publishers, aggregators, and discovery providers that are tenuously connected by software and standards. The other problem is that so much content is behind paywalls. Sci-Hub (illegally) gets around both of these problems by putting nearly everything in a single system without any paywalls.

So it would be great if there could be a legal place one could go that would 1) Contain every journal article and 2) be free of paywalls.

In theory that could be achieved without OA. Like Netflix or Spotify, a third party could hammer out agreements with publishers to provide everything in one interface. Of course, companies are already trying it, and it will never be complete. EBSCOhost, Gale, and ProQuest all offer huge aggregated databases, but those are prone to embargoes and publisher exclusives (much like consumer streaming media services, actually).

And when I say the situation would have to be open, it goes a little further. In order to work 100% of the time, to avoid metadata issues, and to provide appropriate versions, such a system would really need to index the metadata AND store the full text itself. If not, second best would be sending users to a PDF hosted elsewhere, in a sophisticated way that could work around outages, missing files, etc. There’s at least one attempt to do something like that right now, but it’s pretty early days for 1science.

Until then, I guess libraries can work on improving the software and standards that bring this stuff together. In particular, I think library systems need to do a better job of getting users to OA articles (especially in hybrid journals) and offering easy, obvious alternatives when links fail.

And ebooks?

Right, I started with the musing that many of us thinking about ebooks just want them to be as good as journals. Why? Because they are often far, far worse. Ebooks are subject to DRM, logins galore, awful reading software, and poor metadata. Many times I’ve wished for ebook access to be as easy as journal article access, but that would only be elevating books to the level of a still cumbersome system. (Sigh)