In one sentence:
The collapse of a society is a response to declining marginal returns on investment in complexity.
Like all grand historical theories, it's an interesting idea. The downside, of course, is that 'complexity' is never defined precisely, and so the theory remains unprovable. There's good reason why the theory holds for bureacracy and government - Iron Laws of Bureacracy and other perverse incentives - but why for technology?
The thesis is an interesting provocation, but for casual readers, the book (while short) may not be worth the trouble. If you're not interested in proving the theory true or false, this Astral Codex Ten reader review may be a better read.