Zotero, AKA Firefox Scholar, AKA smartFox
Firefox Scholar is finally here. But what do I do with CiteULike?
Zotero (hmm… tastes like some sort of anti-depression drug 
) Beta is released a few days back. It’s touted to be the open source alternative to EndNote and CiteULike. From what I can read about its design and philosophy it is indeed closer to my original vision of how collaborative bibliographic information discovery/management should work.
Some complaints about CiteULike are in order.
In email conversations with Richard Cameron, the creater of CiteUlike, way back when it was first released, I said it should be open-source, distributed (RSS/RDF), and API-aware. Richard disagree with first two. He sees CUL as a service rather than software; releasing the source code, he argues, would undermine the reliability of the service. He also maintains a centralized model, where all data are stored at CUL and are only accessible through CUL. There has been a few server down times in the past years (although overall it’s been amazingly robust) and caused problems.
As to API, there was an effort to encourage open source scraper development, and API was at some point on the priority list, but the fact is years after its release there is no API. As a result many useful features — such as discoving users sharing similar interests, which can be implemented easily with APIs — remain empty promises. And now CUL scrapers start to sease to function as publishers change their HTML structures. There is no way but to email Richard for help.
Unless Richard is planning a CUL2 (PyULike is anything BUT without major rethinking of the CUL philosophy), I fear CUL will eventually see its users jumping to other bandwagons.
And Zotero is potentially one of them. I tried it today. It has a pretty slick interface (which I don’t care about; I think the primary focus of something like this should be developing scrapers, API, collaborations). When it finds a page that it knows how to "scrape", it works fairly well. (A problem, though, is that it doesn’t detect duplications. ) Haven’t had a chance to play with the collaborative functions … for the following reason:
It requires Firefox 2.0, which is not here yet! The FF RC2, which I had to install just for test driving Zotero, is incompatible with the current version of GreaseMonkey (or the other way), and all the GM scripts I wrote for CiteULike won’t work. As my (academic) life depends on CUL for the moment, I had no choice but to uninstall the RC2 after a few minutes of playing at lunch and reinstate the good-o FF 1.5.
Don’t get me wrong. Seeing CUL’s minimalist, boring interface is comforting for me. I have well over a thousand references there, most of which were hand-entered by me. I also developed GreaseMonkey scripts (link) so that I can get PDFs with one click, copy ref from other users with one click, … That’s hundres of hours of work over the year or two.
What do I do with the investment? I am tempted to write a script and download all my CUL data to local, and wait to switch to another system. In fact, with less than a million records, it’s not that hard to down-’em-all. After all, Richard has repeated said that the content in CUL belongs to the public.
But then what? Dump CUL?
I see no reason why CUL and Zotero can’t work together. I don’t know how Zotero’s collaborative server wroks, but conceivably CUL can be one of the repositories. CUL, on the other hand, needs to be RSS-aware, API-suppoting, and Open Source (!). I can do Tcl, Python, Javascript, PHP, you name it. For the hours I spent on GreaseMonkey, I’d much rather help implement it on the server side and make the functionalities available to all CUL users.
This is a plea to Richard: CUL is working well, but its future hinges on its flexibility and openness. It’s monoply won’t last for ever. It’s time for CUL 2.0, one that harnesses user contributions (both in content and code), a distributed infrastructure, and the power of APIs.