blog

The research is done, and all my data is backed up in three different computers (in two different buildings, just in case), two external hard-drives and also in my institute servers.

I also have four chapters of my thesis almost ready, and the latest version is saved on my laptop.

Next thing, disaster strikes, my laptop decided that it had enough and decided to give up, meaning I had lost my manuscript. This all happened within a couple of weeks of having to deliver the first draft.

Fortunately, I had saved it on several pen-drives, emailed it to myself and had a copy on Dropbox; unfortunately, I couldn’t remember which was my most up-to-date copy. After a couple of hours of life-shortening stress, I found a version that I had backed up three days previously. I couldn’t remember everything that I’d lost but on the bright side, I could still finish it on time.

Fast forward 13 years later, and this problem returned to me but from a different perspective. Having spent the last 4 years building F1000Workspace, backing up documents to a reference manager was not something that came up in discussions with researchers. That is, until a user contacted us asking if we automatically uploaded the Word document when using the F1000Workspace Word-plugin to insert citations, as he had lost one full chapter of his thesis when his computer died on him. We couldn’t help this user, unfortunately, but this déjà vu, made me wonder how big of a problem this was and if we could do something to help.

From conversations with researchers, we quickly realised that losing files was more common than I had initially thought. We also unearthed another related issue: keeping track of the different versions of a document that they have emailed to themselves or saved in multiple pen-drives to enable them to work on different devices.

For us, this was an opportunity to solve a problem that we haven’t initially considered, but that was shared by many of our users. Upon exploration, we found that it was not (too) difficult to develop such a feature, as we already offered the functionality of uploading a Word file so it can be shared and feedback gathered when collaborating on writing a manuscript.

In the first iteration, we created the ability to manually upload the word file to F1000Workspace. The file is then kept safe online, together with the previous versions of the same document, and can be downloaded remotely from any computer.

However, even though this solves the problem of dealing with the different versions and having a saved backup of the manuscript, it was still dependent of the user remembering to upload the file every time they work on it.

So, to deal with this, we decided to implement an auto-backup functionality (currently only available for Windows). A user just needs to set it once, and the document continues to be automatically backed up each time it is updated.

To keep the number of backup files manageable and not interfere with the performance of Word, we compromised on the frequency of backing up. We settled for every hour, but only if changes are saved, and we limited the number of backups to 25 in the current week, 7 in the week before, and 1 in each of the previous weeks. Sounds a bit complicated, but we believe that it fulfils the goal of creating regular backups, accessible for everywhere and without storing millions of files.

I hope you will find it useful. Suggestions about this new exciting feature or any others are always welcome. Send your suggestions or any questions to feedback@f1000.com.