Recently a friend of mine posted about losing his entire research library in OneNote, which brought flashbacks to losing my entire research library many years ago in the Endnote 7 to X upgrade.

As such I am using this as an opportunity to briefly review my own research and backup practices, and run through how I use Zotero as a cloud synced, constantly backed up, and ultimately human readable research manager.
Zotero
Firstly, I have posted about Zotero a few times now, so there are more details in those posts. All available here: http://www.porterblepeople.com/tag/zotero/
Zotero functions as the heart of the system and is used to make access to your research library quick and easy. But it is not the only part of the system and requires other aspects to function effectively. As I go into in one of my other posts the Zotero cloud storage capacity leaves a bit to be desired, so I store all of my PDFs in Dropbox. Zotero makes it easy to rename all the files easily and keep soft-links to the database.
Zotero also allows you to keep notes with your documents, and those notes are indexable.

Those notes are also synchronised across devices using the Zotero cloud, and as raw text they take up very little space, so no worries there.
Zotero Export (BibTeX)
However, on its own Zotero isn’t great for being human readable. The last thing you want is for a service to decide that they are wrapping up and then you lose access to all your notes taken in that platform. I have had this happen with old OneNote, and seen it happen plenty of times with other platforms.
Mercifully in the research manager world there is an open-source standard called BibTeX. Now BibTeX is usually used to generate references for LaTeX. If you are a LaTeX user then this is great (and this post is unlikely to be news to you) but for anyone else who doesn’t use LaTeX then this aspect is likely to be somewhat moot. Rather we are interested in the fact that it generates a plain-text human-readable export of your database. In the event of your software being withdrawn or rendered inoperable by something (e.g. 64bit upgrades) then you will always have all of your research entries and notes in plain text.

I will be the first to admit that it isn’t particularly pretty, but it is human readable and that is what matters. Plus it is entirely likely that someone out there will have written a converter to other platforms or software that uses BibTeX as an intermediary step, and therefore you can restore your data somewhere else. Absolute worst outcome is that you can still read your data without having to deal with a closed software format (and you could always learn LaTeX and keep using it).
To do this Zotero needs a little plugin called BetterBibTeX, which I cover in this post: http://www.porterblepeople.com/2015/06/zotero-addons-extensions/ In that plugin you can export your library (File>Export Library…) and just check the “Keep Updated” box on export. I store my library in my Dropbox folder so that it is always cloud synced.
Dropbox
I use Dropbox as my cloud storage of choice, but most cloud storage will work well enough. I quite like Dropbox though as it also offers snapshots and the ability to roll back and restore files. Anything you use for cloud storage should offer those, in case you accidentally delete an entire directory. But if the cloud storage you choose offers the ability to roll back or undelete files, then use that.
Fin
That is it. Expecting more? It is a rather simple and robust backup system for a research library. I have had a hard drive die on my laptop before and it was annoying (downtime) but all of the data was restorable quickly and easily (being on 3G made it a bit slower). I was back up and running again in about 8 hours.
tl;dr? Critical things: keep your files backed up, and your research database exported in a human readable format.
Appendix A
As some friends have pointed out using a cloud service as a backup is only one layer of redundancy, and may not save you. Especially if you have sync conflicts or longer offline periods. However, in this I am aiming for a balance of automation and simplicity, as complexity often introduces barriers to adoption of new practices.
Personally I have Dropbox setup for constant sync which keeps my 3 machines synced, plus Time Machine backing up to a local server, and that doing a nightly rsync to my US server, and I also take a portable HDD into work that is on monthly rotation. Paranoid? No, I worked for a couple of unis in support roles and have seen first hand the devastating effects of losing research too many times. But it is a lot more to setup than a cloud service.

About Chris