That is, in a nutshell, the conclusion of this study. If you have free access to a lot of literature, you are much more likely to click on links and download PDFs (which hopefully means you will read the papers, learn from them, improve your science, and cite them when writing your own manuscripts). If you know that most of the time you will see a "pay $60" page instead, you don't bother clicking anyway.
Also, this mainly applies to the new papers - the older papers are rarely looked at - so there is no real need to keep archives TA for any lengthy periods of time.
Peter Suber comments.
More like this
Douglas Kell: The Matthew effect in Science - citing the most cited:
Shri Kulkarni is the McArthur Professor at Caltech.
He has about 300 refereed publications, and is in the close vicinity of a major numerological birthday anniversary.
Of his papers, about fifty are in Nature!
What happens when I mention a paper describing two more Drosophila genomes?
First, it was anti-vaccine "martyr" Andrew Wakefield's infamous 1998 Lancet paper.