The UN had predicted global mortality rate for the years 2015 to 2020 of 8.1, however based on CIA World Factbook data for 2016, the figure is estimated at 7.78. This was down from 7.89 in 2014. Other figures are available, but the trend is downward.
So the real argument is that 2016 was a big year for celebrity deaths. My opinion is that there were a number of exceptionally notable deaths, which generated a large amount of attention. This gave online content creators a theme to build around. Deaths which would have been notable, but reported much more subtly in previous years, became elements of the larger 2016 story. A kind of confirmation bias was applied. I tried to find figures to support or deny this theory.
I have aggregated the notable deaths per year based on the Wikipedia entries by month, and 2016 does not seem like an exceptional year when you look from 2013 to 2016.
A caveat here is that I have used notable death data. Notability and celebrity are not necessarily equal, however, my entire premise is that a number of notable deaths were considered celebrity deaths, which helped to feed the concept that there was some unusual number of deaths in 2016.
While building charts, I noticed a pattern in the deaths, depending on the time of year. I thought that perhaps this pattern might be related to reporting errors, or an error in the way I extracted the information.
I guess it is quite obvious, but it turns out that this basically reflects the percentage of the number of days in each month. (January has 31 out of 365 days in a year, so 8.49315% of all days in the year are in January — as such, you would expect 8.5%ish of notable deaths to happen in January).
The interesting exceptions on the 10 year notable death curve are December and January which have amplified positions on the curve, compared against the month sizes. May and September seem to have less deaths than you might expect, when compared against the size of their months, and they are the only two months which defy the rises and falls of the "size of the months" curve.
Finally, I am using Wikipedia Data, so my sources and methods are easily brought into question here. I have merely counted the number of deaths reported as notable on Wikipedia for particular days.
This final graph shows how significantly the number of notable deaths has been rising since 2006. It is hard to determine what has caused the steep increase since the mid 2000s, but I think it is fair to at least say that if there is an increase, 2016 was far too late to sound the alarm.