Bypassing the paywall, legally

I've recently come across unpaywall, a web browser plug-in that spots when you are looking at the abstract of an academic article on a scholarly journal's website and adds in a link to a free version of the article, if one is available legally somewhere online. 

Many if not all journals which publish academic articles (and charge for them) allow authors to put copies of the articles (perhaps in some pre-publication format) up on their own websites or on institutional repositories.  At the University of Surrey, we have such a repository, and I thought I'd check to see if unpaywall will pick up articles that we've put there.

So, picking not quite at random an article of mine which is available free from our institutional repository here, I go to the Nuclear Physics A page where my article is published

Lo, with the unpaywall plug-in installed, I see a little green unlocked padlock logo on the right of the page, that is not usually there:

If I click on the logo, I get a pdf copy of my paper from the Surrey ePubs website.  This is handy!  For me, it will be most useful while travelling, since in my office I can download the journal articles through my institution's subscription for journals where they have one, which is most of the journals where I publish.  This is automatically detected through my computer's University IP address.  On the road, I have to rely on some sometimes clumsy methods of getting the University subscription recognised on a per-journal basis.   

Of course, it only works if there is a legal and free version of the paper somewhere online, and unpaywall has found it,  but I'll keep it installed.  The only downside I can see is that it doesn't work on the default browser on the Mac, Safari, so I'll have to switch to Firefox or Chrome.  That's a drag, but I'm not a browser zealot.  I save that for Emacs.

The great Google Citation Purge

Our Vice-Provost for Research and Innovation writes a blog to share thoughts, ideas, and announcements.  I thoroughly approve of his communicating via a blog, of course.  I end up reading what he says more than some other communications from the top that get put on e.g. the University's intranet.  It's a shame that the blog does not allow commenting, as that's an integral part of the ethos and culture of blogs, in my opinion, but okay, it's perhaps not surprising from a blog on a corporately-controlled part of the University website.  At least the thing exists.

Recently, he wrote a pretty extensive post about sharing and advertising one's research.  It is aimed at University of Surrey staff, and so is partly dedicated to our own institutional repository, but its chief recommendation is to use Google Scholar.  I do, and I've been using it for many years.  I posted in 2012 about an annoyance of Google Scholar in that it sometimes seems to forget about citations that it previously knew about.   It has just undergone a such a moment of amnesia.

My most cited paper, as previously judged by Google Scholar is this one.  As I write this post, the commercial subscription citation database ISI Web of Knowledge thinks it's been cited 216 times (making it #3 of all the Phys. Rev. C papers from 2012).  It even has a little trophy logo, calling it a "Highly Cited Paper":

Google Scholar was recently completely on board with this, listing it as my most highly cited paper, but I can't show you a screenshot because it now thinks it is more modestly-cited, at a mere 33 citations:

Based on my past experience, Google Scholar usually knows of many more sources of citation than ISI -- from PhD theses, for example, if they have been posted anywhere online, yet suddenly the number of citations has dropped rapidly.

Anyone know why?