Ridiculous fluff piece.
The author's premise is that taking the S&P dividend yield (which varies to some degree with the overall state of the economy) and adding 3%, you get a variable withdrawal strategy that protects your nest egg from overwithdrawing during downturns.
Here's the primary (glaring, IMO) flaw with the article: "For history we’ll go back 28 years, an interval that is not far from the plausible investment horizon of someone just now retiring. Also, it makes the arithmetic easy."
Sure - if you cherry-pick a specific portion of history that includes a nice secular bull market, you can justify an effectively higher withdrawal rate.
Fundamentally the author misdirects the audience with a lot of detailed numbers, while glossing over the fact that the only time period his strategy was "tested" against (and therefore can be said to be valid for), was the very specific time period from 1988 to today.