tl;dr. If you want to read anything at all, it is the companion page citation management, which offers more up-to-date practical solutions, rather than pointing out problems in a particular software package you probably don’t use and that I haven’t used since 2012.
Citation management is fraught and awful thing that all academics have to do, and no-one ever does well. Or if they do it well, then they aren’t doing it efficiently.
After nearly 2 years of use, I’ve just switched from Third Street Software’s Sente to Zotero. There were some other things I tried out in the middle there, such as Mendeley, Bibdesk, CiteUlike, Connotea and Papers, but a full n:n comparison is not something I have time for.
Reasons I originally chose Sente over Zotero
- First fully-functional iPad app (of all the reference managers) for low-distraction reading of articles
- Supports synchronisation between multiple devices (e.g. my two computers and my iPad although not my Android e-ink reader)
- OSX native application, which seems to suggest it should be fast
- the best tagging navigation of all the citation software.
Reasons I switched to Zotero from Sente
Basically, because while Sente has an impressive feature list, it implements most of the features badly. I’d rather have the slightly under-powered but more solid and flexible interface of Zotero.
- Sente uses its own built in browser for all reference finding. It is an awful browser. It doesn’t integrate with your password manager. It doesn’t support your campus’s proxy bookmarklet. It renders stylesheets wrong. It doesn’t support all your lovely browser extensions. It ignores your system font settings. The competitors, such as CiteUlike and Zotero, get this right. They use browser extensions to supplement your usual browser with handy reference-saving features.
- Sente is bug-riddled, and crashy. I was spending way too much time filing bug reports. While they mostly got fixed, you can only do this so many dozen times before you go from engaged user to angry user.
- Sente is slow. Very slow, both on laptop and iPad. This is alarming, especially on machines with gigabytes of RAM and multicore processors. I have a database with 10 million items in it on my research laptop which can retrieve any one record faster than Sente can retrieve one from my few thousand citations.
- Zotero I thought would be slow, because it is built on Firefox, which is slow. However, it’s downright zippy these days, much faster than Sente, at least with my (3300 items) reference library
- Zotero has an API with which you can both read and write data. So do Connotea and Bibdesk. But Zotero’s API is cleanest, and has an active community around it, and I don’t feel that I am locking my data away forever if I rely upon it. (Of course, you can always try to migrate data around from anything to anything with BibTeX, but if you have URLs in there, or consort with foreigners who dare to have diacritics in their names, this usually runs into trouble). I can use the API to make changes that I couldn’t make manually, without worrying about that parsing nonsense. Since all citation software is, basically, awful, it is even more important that whichever application you choose, it is one that you can get your data out of it when you find a less awful option. Moreover, some other apps, such as Mendeley, already use the Zotero API, so you know that it’s not going to be a community of one playing with it, which it feels like with, e.g. CiteULike.
- Zotero runs on Mac, Windows, Linux, everywhere Firefox runs. It’s sad to lose the iPad (although there is an… ok… 3rd party Zotero app called ZotPad), but I’m getting increasingly concerned about the evilness emanating from the Apple empire and have started to value options that don’t restrict me to their platform.
- But then, who cares about Sente’s iPad app anyway? It seems to be nearly abandoned. It’s just as crashy and slow as it was in 2010 with the same awkward user interface. It doesn’t resume where you left off when it inevitably crashes, so you spend all your time navigating back to where you were.
- Sente’s duplicate detection is OK when it works automatically, but when it doesn’t, it’s awful. There is no easy way to merge articles manually. In fact, if you get a sync conflict, you end up with two copies of the articles, and no easy way to even see what the differences are between the versions, let alone fix them. Zotero does this effortlessly, with an interactive, multi-way merge.
- Zotero also syncs multiple machines, and does it fine.
- Sente’s exported files tend to have parse errors when you load them into
other citation managers, making leaving Sente a chore. Which cements my
determination to do so.
- Moreover, exported files don’t distinguish between author-supplied keywords, and your own, lovingly curated, tagged navigation.
- Sente insists of having horrible font markup squeezed into your innocent data, and tries to preserve fonts when you copy and paste article data. Blech. Just try it. Export your data and see what nightmares Sente is storing in your innocent “Title” fields. When I exported my 8MB citation database XML, 4MB of it was spurious font markup.
- User interface suffers horribly from NIH, being made of custom UI widgets that don’t behave how you expect in a thousand mildly irritating ways. You have to use the mouse to navigate when you’d rather press “tab” to move between fields, keyboard shortcuts don’t work like you expect, if you have your trackpad set up as “tap to click” you can’t select text fields, and many other small, transient bugs and violations of expectations. More than, even, the non-OSX-native Zotero. I’m going to go back to letting my muscle memory do my thinking.