John Regehr's Six Levels of Error
Readers who enjoyed "The Five Types of Technological Standard" will like "Software Bugs and Scientific Progress" by John Regehr.
The best conceivable advice to the computer industry would be:
"Revisit assumptions."
But of course this is just the kind of advice least likely to be taken.
Of course, because they have too many assumptions to revisit them all. They might have been reasonable assumptions initially, but assumptions age badly, so eventually things get to the point where the assumption is worthless, but things still get carried forward due to inertia. This is, arguably, the fate of any assumption in computing (and probably anywhere else).