Personally, I think Joel has shared some good lessons and insights. However I also think he ran out of really good material about 2 years ago, yet various outlets "demand" more from him.
This tends to happen a lot with almost any person who tries to blog/write too much on any sub-topic. At some point you've exhausted all of your unique datapoints and end up resorting to trying to spin the obvious into new and interesting stories. This is why, among other reasons, I like PG's almost-monthly essay installment schedule. I never feel like he is straining to product content.
It's not about the most efficient way of cleaning toilets or putting up drapes, it's about psychology.
I've managed hackers in different startups, and my experience is that they aren't that different from anyone else in this respect. The primary difference is that they think these rules don't apply to them, which of course makes it much easier to apply them.
I don't see how taking 2 days and leadership time demonstrates concern for employees. Getting someone in who can take care of the problem quickly and efficiently* seems like it's better 'staying out of the programmer's way' than taking two days.
I think the lesson from the toilet cleaning was, "Look, if I'm going to require perfect inspections, I'm sure as hell going to have an immaculate bathroom. I also recognize that I need to show you how to perform to the level I expect. So here's how it's done."
That hands-on, know-what-you're-doing leadership is what's respectable. Making an object lesson, on the other hand, is hardly laudable.
*This falls into the "why I hesitate": if Joel could do it faster and more cheaply than having someone else come in that knows what they're doing, my criticism is less harsh, though I still think he's taking the 'wrong' lesson from his military experience.