In talking to one of my students, I realized that my way of organizing drafts of documents might help others.  Coming from a software background, the only thing that kept me from using an open-source source code management system was that I didn’t want to spend the time setting the tool up.  Instead, I fell back on file naming conventions and backup schemes that ensured nothing could be lost.

1. Create a folder structure first: You should have one for the document itself, one for pdf”s of literature you may reference, one for your data, syntax and output.  At it’s simplest, you can have something like this:

• Dissertation
  • Chapters
  • Data
  • Reference

1. Break the document down by chapter:  I broke my dissertation down by chapter so that it would be easier to edit.  At least twice I have seen a document in word implode due to size combined with track changes and code fields for citations.  Working on each chapter in a separate document let me avoid that problem.  Because I use Zotero, when I copy and paste the text into a single document I can just refresh the bibliography and Zotero merges it all into one.  (Most other reference management software should do the same.)
2. Set your word processor autosave to ridiculously frequent:Seriously.  I set word to save autorecover information ever 2 minutes.
3. When you first open a document on a particular day, save it with a different name: Saving with a new name should be the very FIRST thing you do.  That way you can be sure that the old version is preserved.
4. Create a file naming convention and stick to it: There are lots of different ways to do this.  Here’s mine: 20110228-Ch5-Results.docx
   1. The first string of numbers is the date.  That is the part that changes every day I work on it.  Notice that the year comes first, then the month, then the day.  This ensures that it sorts in order of when the document was created.
   2. The second part tells me where in the overall document this chapter belongs. I added this after I realized that the chapter names didn’t sort into the correct order.
   3. The last section is the chapter name, more so that my husband and my editor can easily find things.  It isn’t technically necessary.
   4. When someone reviews it, they tack their initials on to the end of the document name so as to separate it from the original document.
5. Every time you pause, hit the save button:  More is better.  Or set yourself an alarm to hit the save button approximately every 15 minutes.
6. Set up background automatic copies to a different disk drive: I use Syncback free to copy the entire documents directory to my Dropbox folder, which makes an offsite backup copy.  If you are worried about security, Syncback free can save to a zip file and add a password, or Syncback SE ($34.95) can encrypt the file.  Set it to backup FREQUENTLY (I used 5 minutes), but only worry about the chapter and data folders.  That should keep you under the free 2G dropbox limit.
7. Your data analysis output needs to be kept as well: As important as your writing is the syntax you are running in whatever analysis tool you use and the output of that tool.  Use a similar naming standard, and name the syntax that created the output and the output itself the same.  Try to give the files a name that tells you what you were trying to do that day.  (“Analysis” isn’t good enough…)
8. Clean up the old files each time a milestone passes:When my proposal was approved, I got rid of all the earlier versions of those chapters and saved a set with the names 20110104-Ch1-Intro-APPROVED.docx.  Then when I started working again, I used those versions as the base.  I will do the same thing when I send it all off to the editor, again when I submit, and then with the final version.

The key is to automate what you can (backup copies, off-site backups using dropbox), save early and save often.  I have heard too many horror stories of crashed laptops with no backup copies or bad word documents that were the only version.  All of these can be avoided with a bit of planning and a pinch of paranoia.