This morning, the number of citation styles in the Citation Style Language (CSL) repository went above 6000. This is an amazing success story for a tiny, unfunded open source project. Here are three reasons you should be excited about this.
1. Publishers are seeing the light on citation styles
When Rintze Zelle posted this news on twitter, one response was bewilderment (even anger) about the number of citation styles:
Mike Taylor (@MikeTaylor) April 30, 2013
Chris Rusbridge (@carusb) April 30, 2013
Is it publishing added value if T&F has 500 CSL styles?—
Chris Rusbridge (@carusb) April 30, 2013
But these folks get it all wrong. The reason there are more CSL styles is that there are fewer citation styles. Take a look at this graph, by Carles Pina, one of Mendeley’s developers (not embedded since it updates daily). See those jumps of the green line in mid-April and mid-March? Those were the dates citation styles for a massive number of journals by Springer and Elsevier were added. But now look at the blue line – the number of “unique” styles, i.e. citation styles that are actually different. It hardly budges on those dates. The reason for that is that both Springer and Elsevier have started to consolidate the citation styles for all of their journals. Once they’re done, (almost) every one of their journals will be based on one of six citation styles. The list with those styles was provided to Papers and Mendeley developers when their companies were bought by Springer and Elsevier respectively. Taylor & Francis is in a similar process of consolidating their journals to five citation styles and they just gave us the list when we asked (and were very nice & helpful in the whole process, I’ll add). All three of these publishers base their styles on popular existing manuals such as APA and Chicago. Sage seems to be doing the same — they have style manuals for a Harvard and a Vancouver style — but hasn’t replied to my request for a list of journals following them. Sadly, I haven’t seen any similar movements from the major university presses — Cambridge, Oxford, I’m looking at you.
But why do we add those journals to CSL then? And do we keep 500 versions of the same style? As for the first question – our vision is that if you want to publish in a journal you just look it up in the list of styles for you (CSL-using) reference manager and, voilà, you have correctly formatted references. No need to check “Instructions for Authors” on the journal’s webpage or to figure out if the style they provide is “Elsevier Harvard” or not. And of course we don’t keep 500 copies of the same style on the CSL repository. Where one style is used by multiple journals, we create a dependent style, which is typically about 15 lines long and mainly consists of the name of the journal and a link to its “parent” style. You can see an example here if you’re curious.
2. CSL increases competition among reference managers…
…and that improves the quality of the existing products. Seven years ago, when Zotero got started, and CSL was still in its kids shoes, providing a citation function was a major obstacle for a new reference manager. Zotero invested a fair amount of initial resources getting CSL of the ground, writing the first CSL processor (the code that converts data and citation styles into formatted citations) and contributing a fair number of citation styles.
Today, when a new product like Docear or Colwiz enters the same market, things are radically different. They can freely use many thousands of CSL styles (each one under an open CC-BY-SA license). They don’t even have to write their own processor, but can pick from various options: citeproc-js by Frank Bennett (used e.g. in Zotero, Mendeley), citeproc-hs by Andrea Rosato (used in the wonderful Pandoc) as well as citeproc-ruby (Sylvester Keil) and citeproc-php, (Ron Jerome) both of which are optimized for web-based implementations. They can direct their users to the visual CSL editor developed by Steve Ridout in a cooperation between Columbia University and Mendeley. Thanks to CSL, implementing a citation function in a reference manager has become relatively easy.
In other words the “barriers to entry” in the reference manager market have been lowered significantly by CSL to the benefit of us, the users. More competition doesn’t just mean more choice, it also means more rapid innovation. Even if you’re still an Endnote user (and you really shouldn’t be) you’ll benefit as Endnote is reacting to the increasingly fierce competition by improving exactly those feature where e.g. Zotero and Mendeley excel (or why did you think there is now a retrieve metadata feature for PDFs and better support for webpages?) and improving theirfree offerings (their #freeEndNote hashtag sounds more like a plea to me, though).
3. Free and Open Source Solutions Work
While much of the work on CSL is performed by a small team of unpaid volunteers (essentially Rintze, me, and the authors of the processors), we have been increasingly successful attracting outside contributions. According to the stats collected by ohloh.net, CSL as a “very large development team.” While I have been critical of the lack of support for CSL by Mendeley et al in the past (and still the vast share of new styles and improvements come from Zotero and its users), both Mendeley and Papers have taken a very welcome more pro-active role in CSL recently. Papers developer Charles Parnot has been instrumental in engineering the push to incorporate more dependent styles (and has developed much of the work-flow we use for that). Mendeley contributed the long awaited visual editor (which is open source) and Carles Pina has helped out with the large Elsevier and Taylor&Francis additions. The brand new CSL logo you see up top was donated by Austrian graphic designer Johannes Krtek. So helping hands all around.
All that said, CSL’s enormous popularity does pose some challenges, as the day-to-day maintenance of the repository at times takes up more time than Rintze and I can reasonably muster. CSL currently has no funding of any kind. We’re working on finding ways to make the project long-term sustainable and assure continuing improvements (and I’m somewhat confident that the large number of stake-holder will help in that process), but that’s a topic for another day.