If all the reader gets out of the article is that "systems never perform at 100% efficiency" then they are ignoring the entire point of the article. As such, the root post seems to be hijacking the topic rather than talking about what Schneier actually said.
There is some (insufficient, at least in the US) help available from government programs and whatnot, but even non-participation in such things is not always necessarily by choice.
I've known a fair number of people who've worked their way out of poverty (my family among them), and I've known a larger number who have failed to do so by no real fault of their own. Your unkind generalisations are most definitely unwelcome.
1) In any system of nominally co-operative elements, with ability for individual elements to flip between co-operating or defecting (to use the same game theory terms Schneir did) will have some fraction of those elements in 'defecting' mode.
2) He accepts as fact that perfect control over the choice of defecting or co-operating is impossible to achieve in the systems he's talking about.
3) He outlines a way to reason about the processes and measurements one can put into a system which maximizes for 'cooperation'.
Then Schneir applies this very good and universal tenet toward security. and finally points out that even if systems could achieve perfect control that would be net negative as the 'defectors' in the system can provide the mutation/change which improves the system overall.
It is a solid piece of reasoning which, in my opinon, applies to all systems, not just security systems. Systems that include groups of people, computer networks, air traffic control, commodity markets, Etc. All of them.
Take a Gaussian distribution and point at one of the wings and there be the 'dishonest minority' there. If you are a designing a system where the actors (or agents) in your system have the freedom to choose their own path through it, whether it is an MMORPG or a botnet or a homeless shelter, you have to consider (and prepare for) the fact that a portion of your system's population will be working against the system as a whole and if you don't want the whole system to collapse you need to heed Schneir's advice and layer in processes which 'self heal' or 'detect and deter' such defections. Otherwise your system will become unstable, and then collapse.