Increases are not uniform in two ways; as a fixed percentage of your PIA and as a percentage of your current benefit at the moment you consider "take or wait." The other issue is that being alive one more year, doesn't increase your life expectancy by one year, and the shortfall is not uniform.
Here are some numbers from my post from 5 days ago (where I mention 64 and 67 as the sweet spots - recalled from about 10 years ago)
male age 62 + 19.63 = 81.63
male age 70 + 14.21 = 84.21
Now let's fill in age 66; if everything was uniform we would get 66 + 16.92 = 82.92. The facts:
male age 66 + 16.79 = 82.79.
Surviving from 62 to 66 gets you 1.16 years; surviving from 66 to 70 gets you 1.42 years.
As I recall, and I haven't done any real serious thinking about this in a number of years, 64 and 67 were two of my favorite years. So men would come out slightly ahead by starting at 64 (but only in some probabilistic expected value sense) and women would come out slightly ahead starting at 67.
Of course the biggest issue is sex, which the article never mentioned (I can't believe that the authors didn't).