Rumsfeld was referring to a way of categorizing uncertainties in a mutually exclusive and collectively exhaustive way. There are many scenarios that are genuinely unknowable - it has nothing to do with competence. He probably saw the matrix on a McKinsey consultants slide earlier that day while they were using their 6 months of strategy consulting experience out of college to help him run the war.

The Rumsfeld formulation itself dates to at least the 1950s and development of the Johari window (named for Joseph Luft and Harrington Ingham).

I'd first encountered it in print in a 1982 book, Forced Options, by Roger Lincoln Shinn, two decades before Rumsfeld's famous 2002 usage:

<https://archive.org/details/forcedoptionssoc0000shin_c8x7/pa...>

Rumsfeld's embarrassment over that line was well earned. He was trying to sell a war based on Iraq having weapons of mass destruction. He was asked why we should go along when there was no evidence for that. He tried to sound smart saying, "Sometimes you don't know what you don't know", basically admitting that there was no evidence to justify his predetermined course of action.

They weren't "unknown unknowns". We knew. He knew too, but he was trying to muddy the waters. He got his way at the expense of one million dead Iraqi civilians, thousands of dead US soldiers, and 3 trillion dollars of US money.

No, it is much closer to the Cynefin framework's understanding that complex systems are affected by the influence of variables which are too small for us to measure or even know they exist until after they act.