Are APIs Enough?
Martin Fenner discusses issues related to a recent post of mine, and concludes, first:
Both Microsoft Word and OpenOffice should open up their citation APIs to third-party tools. This would create better citation tools, allows the easier exchange of documents between authors (journal submissions), and would it make easier for smaller tools such as Papers to integrate with word processors (a workaround is described here).
I guess it depends on exactly how the APIs work. My view is that they probably need to allow users to easily connect different kinds of data stores, which then send back the relevant metadata, which gets embedded in the document in a standard way. Citation processing, then, probably needs to happen closer to the document than the bibliographic plug-in (or whatever).
The citation processing also likely needs to be pluggable. Microsoft’s XSLT-based solution, for example, is not very good. OTOH, if they were, say, to write a fast C# or F#-based CSL processor that fully supported the spec, that problem would go away.
Second, he warns:
If they wait too long, we will probably see the online word processors such as Google Docs, Zoho Writer start adding an API for citations and bibliographies and all of the sudden become very serious alternatives for writing scientific papers.
I’m not holding my breath, given all the other limitations and problems I see in Google Docs. And if they add it, they still need to consider the document interoperability concerns I note above and in the earlier post. But coupling APIs with standardized and embedded document metadata can help solve all of these problems.
Thanks for picking up the discussion from my recent blog post. From the current limitations of citations in Microsoft Word 2007/2008 (citations stored in one flat file, limited citation styles) I see the first one as more important. Allowing access to different kinds of data storage (and a search function) would go a long way. Citation styles are less important, at least for journal submissions. I do believe that we have far too many citation styles, and journal submission systems usually check and transform the bibliographies anyway.
The second best bet would be a standardized plugin that is supported by more than one reference manager.
Hi Martin: I would have posted a comment on your post, but didn’t feel like signing up.
It’s funny/interesting that you should say citation styles aren’t that important given the extreme anxiety often on view among academics when an application like Zotero gets a minor detail wrong. It’s unfortunate, really. But I do think in general it is a real issue (notwithstanding some exceptions).
I see efforts like CSL as a way to drag this stuff into the 21st century. I would hope that over time, though, the academy will be able to rationalize this and drop the myriad of archaic ways that we communicate something pretty basic: how to find a source. The web really should, and can, change how we do this.
But, as a friend of mine once said, the academy is more conservative than the Vatican!
You are of course right about academics being rather conservative. And CSL is a great effort. But in a citation I only care that it finds the source, and at least for academic papers we have the Digital Object Identifier (DOI) for that. I would rather leave it to the journal - which is cross-checking the citations anyway - to format the bibliography in the style they prefer.
To increase readability of the draft manuscript, you probably want to add some basic info to the DOI, e.g. first author and year, but 1000+ citation styles?
Martin, I guess what I’m saying is that in my experience (in the social sciences and humanities), this …
… doesn’t really happen, and I’ve heard of people getting major grief (from editors, advisors), etc. about minor formatting mistakes. Consider also fields like history, where people are citing 15th century manuscripts. Suffice to DOIs are uncommon there. E.g. I agree with your priorities, but just don’t think we’re there yet.