Here is Matthew Cockerill's comment on http://poeticeconomics.blogspot.com/2008/04/for-sale-cell-wall-structural-changes.html
(posted with permission):
Anyway, this isn't problematic and it is not a theoretical possibility,
it is already happening in the real world.
There are plenty of service providers (including Infotrieve and the
British Library) who will, if requested deliver a printed copy (or an
electronic copy, for that matter) of an OA article, on payment of a
service fee (but not a copyright fee).
From BioMed Central's point of view, this is a good thing. There are a
fair number of information professionals, whether working at pharma
companies or in the academic sector, who value the convenience and
time-saving of being able to go to a single source to order up a range
of documents (both OA and non-OA), and to have them all promptly
delivered as a collection. Both Infotrieve and the BL will do this,
charging users a service fee to do so.
A "no commercial use" constraint would prevent this kind of
redistribution, and so lessen the ease of availability of OA articles in
PS obviously, if the wording suggested that the service provider *owned*
the research article concerned, or that the user *must* pay a fee in
order to read it, or that the fee concerned was a copyright fee, then
that would be false advertising.
Heather's comment: a document delivery fee-for-service (as charged by Infotrieve and the British Library) is different from charging for the content per se. There has to be a way to enable this kind of distribution, with a blanket commercial license. More later...