Here's my corrected plot of days in the bottom 3 for each year (lesson: more haste, less speed).
For each date I've calculated the actual ranking to be in the bottom 3; this is needed because of data gaps (eg late July-early August 2002), which throws all the other years' rankings by one place. So I've definitely identified the bottom 3 years for each date now. This gives the correct cumulative plot, attached. I've excluded years that by Dec 31st had accumulated fewer than 20 bottom 3 days for any date.
Thanks to all who responded to my first, rather excitable post on this one - you were sceptical in the proper sense of the word.
Looking at the next 10 days: 2005, 2006, 2007, 2015 and 2016 all have days that are currently 25th out of 27 (see excerpt of rankings attached from my .xls). In the likely event that 2017 extent remains in the bottom 3, its cumulative score will gain one each day, and one of those other years will lose one each time.
The inevitable question is "so what?" Does this (in March) really matter when it comes to the showdown - the summer minimum? I don't really know the answer to that, but my hunch is that it might say something about the overall condition of the ice.
And don't worry, I won't post overly regular updates of the chart! It could get a bit tedious. But maybe monthly...