But the one critical aspect of privacy that Posner misses is that it can be used to hide dissent. And that is crucially important to any free, liberal society.
Political freedom is meaningless without the freedom to conspire for a change of government, in favor of opposition candidates, for changes in laws, in general to work to overturn the powers-that-be...without those powers-that-be knowing about it.
I'm with David Brin in thinking that privacy is, in general, overrated - so long as officials and those in power are willing to be equally transparent (which they aren't).
But this one crucial aspect of privacy is absolutely critical to any semblance of democracy.
They didn't say to use C and C++ for mission-critical systems, they said to learn them.
Learning low-level, primitive languages (including assembly) teaches valuable skills that are applicable in many other areas. Every well-seasoned programmer should know some of these low-level languages - well.
That said, despite the fact that you're correct that these languages have little or nothing in the way of "safety features", they nonetheless have a critical role to play in the software ecosystem. Almost all "high level" and "safe" languages today are implemented in C or C++, and for very good reasons (having to do with memory efficiency and fast execution).
As such, they are not "among the very worst in existence" - they merely have strengths and weaknesses, and places where they're appropriate, and not appropriate.
Just like all the other languages.
[FWIW, I have written a great deal of real-time code that runs close to the metal. Some of it even has flown in rockets.]
The problem is that sometimes a patent covers a real breakthrough technology that vastly improves the functionality or performance of standards-conformant products.
Then you have a dilemma. You can follow your rule, and just not get those gains (in an interoperable standard) for 25+ years, or you can somehow deal with the patent.
In some cases the patent holder might be willing to license for free in exchange for incorporating their technology into the standard. But not always.
In those case you either need to expropriate the patent holder (assuming that the patent is legit, which only 0.1% are - but that's another rant), or you need to have a system to compensate the patent holder without allowing them to (a) bamboozle the committee re what the license terms will be or (b) play favorites among the licensees.
Patentable inventions are more-or-less inevitable. If person A doesn't invent a thing, eventually Person B will invent the same thing. Maybe 6 months later (if it's a bad patent), maybe 20 years later (if it's a good patent). But eventually.
Copyrightable works (books, music, etc.) are not inevitable. If The Rolling Stones didn't record Satisfaction, nobody ever will. If Neal Stephenson didn't write Cryptonomicon, then nobody ever will. Ever.
The (supposed) benefit to society from patents is to accelerate technological progress, by making inventions public sooner. [Our present implementation seems to instead make things worse, but that's the idea behind it.]
The (supposed) benefit to society from copyrights is to encourage authors and artists to create, by giving them a way to profit from that creation. This also doesn't work very well anymore (we need a new method, yet it did work OK before computers made things easy to copy). But that's the idea.
I don't support the status quo implementation of either patent or copyright law - a reasonably good idea has become perverted - but there is a very good rational reason why copyrights are longer than patents.