Thinking about it more, I realize that it’s not such a huge problem … while in conversation scientists often refer to “PNAS”, “JBC”, and “JMB” (PubMed gets JMB ‘wrong’, and won’t understand JBC at all), but if I actually look at reference lists in journals, those abbreviations are never used. It is always “J. Biol. Chem.”, “J. Mol. Biol.” and “Proc. Natl. Acad. Sci. U.S.A.”. Not sure where I got it into my head that “JMB” was actually used in any real reference list.

The key “user story” I envisage for ResolveRef is something like:
“Dr X. has a printed journal article (or pdf) and quickly wants to link to an article in the reference list, then post that link to her blog. She quickly types the key details (journal/year/volume/page) directly into her blog (with the ResolveRef domain in front), presses “Preview” and tests the link. It forwards to the article as expected. Dr X. saves making a trip to HubMed (or maybe PubMed, or GoPubMed), and keeps working on her blog post”. So as long as the user has the reference in front of them to copy, guessing at abbreviations like “JMB” shouldn’t be a big issue.