While I can see your point that story points should never be thought of as a unit of time, you later reference measuring sprint velocity (amount of points burned per sprint).

What's the purpose of sprint velocity if it can't be related to time?

Following the thesis of the article, if I filled a sprint full of 0.5 point stories in one sprint and then 3 point stories in the next, wouldn't the velocity be off by a factor of 800%?

Isn't that a demonstration of the inaccuracies using story points encourage? And how secondary concepts like sprint velocity make the basic assumption that is incorrect?

You could assume a normal distribution of story sizes in relation to velocity which would undoubtably be true over a long period of time, but why both with the inaccuracy to begin with? Why not just measure velocity in units of distinct story sizes?