Re: This will happen again.
[apologies for late arrival; although a regular reader of TP's stuff, I missed this one till an occasional visit to Ross Anderson sent me here from
"Is this where we are urged to use Linux instead of Windows?"
I don't know about that, but it might be a bloody good place for people to start thinking (maybe even thinking again, for some people) about building stuff that is based on open and non-proprietary standards and interfaces (APIs, protocols, whatever) rather than today's default proprietary implementations such as Win32, x86, whatever. The ability to change implementation platform without too much hassle does have some value, though it's hard to put it on a PHB's spreadsheet.
If telephone networks had to be built on implementations rather than on standards, like computer systems allegedly have to be built on implementations rather than standards, telephones would still be in the age of dialup, and companies like TalkTalk and Freeserve would never have existed (hmmmm). Try the same line of thought with electricity supply. It's not pretty is it.
Plenty of similar examples, e.g. cars in recent decades have been built around standardised interfaces not standard implementations, though we seem to be going backwards somewhat in the last few years in that respect.
There have been lots of these various non-proprietary technology standards around for many decades, and until there was a de facto OS monopoly in the IT Manager's heads, it was reasonably possible to write quite complex code (not just "hello world") and build quite complex systems that could relatively simply be moved from one platform to another, so long as the relevant standards, interfaces and protocols were available (and behaved largely the same way).
The world of popular RFCs is one such example. As might be (for example) a dusty deck of FORTRAN code. FORTRAN is FORTRAN, right?
POSIX attempted to formalise this concept somewhat, as did Open Systems Interconnection (wow, a networking abstraction that accepts that there is more to the world than USASCII and two potential byte orderings for integers, and that standardised interfaces to multiple authentication mechanisms might one day be useful, what a set of concepts).
The successful examples above are *engineering* examples. But IT isn't engineering. IT has somehow become a fashion-driven industry, and in particular a fashion industry where not just one size fits all, but one brand fits all too, courtesy of the PHBs in the IT Manglers office, with their spreadsheet-driven MS monoculture.
Wise people, even those without MBAs, knew there'd be a price to pay sooner or later. And here we are, just as the wise people predicted, lots of innocent people are paying the price of "shiny but defective". Some people are paying a very big price.