“SFX Miscellaneous Free Ejournals Target: Usage Survey Among the SFX Community“, Serials Review (2015), 41(2), pp. 58-68.

SFX is an OpenURL link resolver product for university libraries, focussed on the output of traditional publishers — of which 16-20% is apparently so dodgy in terms of quality that it breaks the system. Yet, rather amazingly, it appears that much of this 16-20% is still allowed to get to the point-of-use.

The article briefly surveys recent findings on how SFX copes with open access articles, and then the rest of the paper gives the results of a survey of librarians who integrate a specific ‘free’ section of SFX with their library discovery tools. It appears that scholars looking for open free full-text via SFX can expect way over 20% dead link errors on URLs…

… one category [of failure] (incorrect parse params) alone leads to 20% false positives (dead links) for MFE [the largest ‘free’ target in SFX]. Besides incorrect parse params, there are numerous other reasons for the occurrence of false positives (dead links), such as resolver translation error, inaccurate embargo data, provider target URL translation error, incomplete provider content, wrong coverage dates, indexed-only titles mistakenly considered as fulltext titles, and other reasons listed in the literature review section.”

So that might mean… perhaps 40% of links to open access full-text are dead? Or even more, like… 60%? The article doesn’t hazard a guess.

The DOAJ ‘targets’ are apparently not much better…

It’s an irony that I find discovery services generally have much poorer coverage of Open Access than Google Scholar. … Most discovery services have indexed DOAJ (Directory of Open Access Journals), but many libraries experience such a bad linking experience they just turn it off” — Aaron Tay, July 2015.

I’m pleased to say that JURN should have close to zero dead links on standalone journals, due to the way it is set up. JURN may lead to a few fleeting “server maintainance” / “timeout” errors here and there, but if the journal’s base URL for articles moves then its articles effectively get auto-removed from JURN’s results. But they get found again within a year at most, through an effective two-pronged method.