+1 on Sean Carroll. I think he overestimates the strength of human cognition, but that's not so vile sin as underestimating it. And he's a very good speaker I terms of getting his point across.

But yeah. There are definitely entropy analogies in software development. A couple of places I've worked at I've described as "brownian." Engineering would move forward a bit then marketing would change the requirements and we would make progress towards the new requirements, but before long we would get new requirements. If you mapped out progress starting at the origin and then put different objectives along the peremiter of a unit circle, progress would look like the random walk of an atom in a warm gas.

Those are the places where you ship a product when progress accidentally meets expectation.