In the following, I describe the results of a brief experiment with the publishers whose publications I access most often. The publishers in the test were Elsevier, Springer, Wiley, Taylor & Francis, World Scientific (WS), IEEE, ACM, American Physical Society (APS), SIAM, Royal Society Publishing, Nature Publishing Group (NPG), Science/AAAS, and Proceedings of the National Academy of Sciences (PNAS). I also included Google Scholar (GS), as unfortunately I use it frequently for finding citations.
The purpose was not to access papers: only to obtain information about the papers. Hence I tested two functionalities: accessing the abstract page and exporting the citation to Bibtex. The procedure was to search for the title of a paper published by one of the listed publishers with Startpage. A click from there would take me to the abstract page. From there, I tried exporting the citation. With Google Scholar, the search took place in Google Scholar itself.
The Japanese and the Swedish IP addresses did not make a difference in the outcome, hence the two cases are not separated. The blocking of Tor exit nodes is inconsistent, hence the results may not be easy to reproduce. In any case, if Tor worked, exporting the Bibtex citation was also possible. The results are summarized in the following table.
Taylor & Francis requires cookies. A warning comes up if cookies are disabled, blocking you from further access. Among other nonsense, it says "To provide access without cookies would require the site to create a new session for every page you visit, which slows the system down to an unacceptable level." SIAM and World Scientific will tell you the exact same message.
Not all Tor exit nodes are blocked systematically yet. I noticed that from certain exit nodes, Royal Society Publishing and ScienceDirect are blocked: the pages would not load. I imagine attacks are frequent from Tor exit nodes, and they choose the most brainless form of defence: block the IP address. Their profits are obviously not channelled towards hiring competent IT personnel. I am not surprised to see ACM on the list of blockers: they are sponsored by the NSA, they glorify the actions of the NSA by calling the employees `law-abiding dedicated patriots' in their flagship publication, their new spam filter is even more closely tied to Google, and they recently added tracking links to their newsletters. Blocking Tor exit nodes to access the digital library is the least they can do.
The most hilarious finding of this experiment is that Nature only exports citation to RIS format. It must be hard to provide a Bibtex option.
We all know that academic publishers are evil. Until now, they only wanted to cash in. They are going a step further and they are ramping up efforts to harvest data about you. For what? I prefer not knowing. Most of the major publishers are either based in Five Eyes countries, or have extensive operations there. Espionage is in their blood. The state of affairs is depressing as it is, and the situation is only getting worse.