The conclusion looks correct for the wrong question: isn't this the formula for the number of queues?
The first car starts a queue with probability 1, the second car starts a queue if and only if it is slower (probability 1/2), the third car starts a queue if and only if it is the slowest so far (probability 1/3), and so on. Total is 1 + 1/2 + 1/3... which is the formula at the end of the blog post, with an off-by-one error.
The average queue length should be the number of cars divided by this harmonic sum. Which also diverges to infinity.
The first car starts a queue with probability 1, the second car starts a queue if and only if it is slower (probability 1/2), the third car starts a queue if and only if it is the slowest so far (probability 1/3), and so on. Total is 1 + 1/2 + 1/3... which is the formula at the end of the blog post, with an off-by-one error.
The average queue length should be the number of cars divided by this harmonic sum. Which also diverges to infinity.