His work taught me to watch out for such problems. Plus, keep a mix of formal and empirical methods on any given project so they cross-check each other.
Extremely interesting to read how data availability is very often (in the military, no less) more useful than both confidentiality (because any eavesdroppers are unlikely to be able to act on the information quickly enough) and authenticity. I wonder if the latter is because of some combination of the difficulty of crafting a believable message which will have the intended effect and the hardware requirements, because the sky isn't full of nuclear missiles.
The part on the placebo SSL certificate is interesting (https://www.cs.auckland.ac.nz/~pgut001/pubs/book.pdf#page37). Do users really ignore the big red warnings that web browsers give for invalid certificates? I imagine that it has to do with the fact that someone would rather be able to do what they want to do instead of heeding a warning they don’t understand.
On further reading it seems like just the intro section is (imo) unnecessarily hostile. This is a pretty engrossing read - although I'm skimming through and just reading all of the amusing anecdotes.
Specific example: Phrases like "Fashion Statement Technology" seemed kind of unhelpful, better keeping the explanation of why they are bad without the snarky labels.
https://web.archive.org/web/20170214231046/http://www.cypher...
His work taught me to watch out for such problems. Plus, keep a mix of formal and empirical methods on any given project so they cross-check each other.