In the entry below, I was not sufficiently careful to avoid Nature-bashing, or the implication that Maxine Clarke was morphing, werewolf-like, into some kind of publisher pitbull. Thanks to Pedro, bdf and RPM for responses which made this clear.

Peter Suber provides a handy roundup of Nature's OA and free-to-read offerings:

[the Current Science partnership] won't be Nature's first OA journal.  Nature and EMBO publish Molecular Systems Biology, a full OA journal, along with a couple of hybrid OA journals.  Nature publishes another hybrid with the British Pharmacological Society.  It publishes a regular series of OA supplements to its flagship TA journal, and in January of this year began offering OA to the backfiles of its academic and society journals. 

In addition, Nature has a raft of non-journal OA projects, including a self-archiving policy, a data sharing policy, a neuroscience gateway, a signaling gateway, a networking site, mixed journalism and research sites on climate change and stem cells, blogs, podcasts, job listings, a news aggregator, and a preprint exchange. 

[Updated after talking to Timo Hannay to include] The Cell Migration Gateway, Dissect Medicine, The Functional Glycomics Gateway, GI Motility Online and The Pathway Interaction Database

It's worth noting that Peter uses the term OA for services and projects which I would describe as free-to-read (or free-to-use), but not OA. I would welcome clarification from Peter here, as I do not feel I am in a position to argue OA definitions with someone who helped draft its founding declarations! [update: see comments]

Even on my more restrictive reading, Nature does have a couple of full-OA journals and a handful of hybrids -- not "one barely-OA journal". Further, whether or not one considers them OA the free-to-read/use projects and services include some important and useful innovations. (The list above doesn't even include Connotea, a science-centric social bookmark manager which I use myself.) Nature is head and shoulders above any of its toll-access competitors in terms of web savvy and willingness to experiment, and I think it's important to recognize this whenever one (quite rightly!) criticizes them for not (yet) being Open Access.

What bothers me about calling Nature's free-to-read/use publications and doohickeys "OA" is the Low Fat/Greenwashing Problem, which Peter Murray-Rust describes thus:

Publishers blaze around "free" "choice", etc. which confuse rather than inform. For a publisher "open" and "free" are to be used like "low fat" "energy food" "healthy" as a way of legitimising current practice.

Everyone is familiar with companies which label their products "environmentally sound" or "healthy choice" when in fact they are paying only underhanded lip service to those concepts. It seems to me entirely possible that unscrupulous publishers may try the same tricks with "open access", and that the best defense is to insist on the BBB definitions. A number of commenters have wondered (can't find a link right now) whether we need another term for Open Access sensu stricto -- something like "BBB-OA", perhaps. (If you say that "be-three-oh-ay" it's not so bad.)

Let me finish, though, by pointing out that I do not wish to paint NPG as one of the unscrupulous publishers whose intentions worry me, nor Maxine Clarke as their sneaky shill. If NPG uses the term "open access" differently from me, I take that as a good-faith disagreement, and if Maxine uses the term in her employers' sense that is hardly "marketing". Specifically, I apologize for the phrase "if [Maxine] is going to start abusing [the term "OA"] as marketing for Nature", which contains an uncalled-for implication that I hope this entry will dispel.

You can get to like the taste of crow... you just have to eat enough of it...