Thanks Jesse! I’m no social-scientist, but here’s a few ideas why the time trend is so large:

• Increasing use of Wikipedia. It was still tiny in 2002.
• Increasing use of the internet and mobile devices.
• The internet now gives us immediate non-stop celebrity-news, so we can update their Wikipedia bios constantly.
• The number of celebrities has increased, due to the growth in media and new ways to be famous.

I haven’t looked into whether edit activity does greatly decline after death. That may be a large source of inaccuracy — I should have investigated. I believe that the linear interpolation mostly compensates for any trends in the area, so the general conclusion (‘2016 was a deadly year’) remain true.