2 ms·
I find weird and worrying that you cannot analyze the safety of a program easily with some form of static analysis, or that there is no solution to do an automa
by jokoon 9y ago
I find weird and worrying that you cannot analyze the safety of a program easily with some form of static analysis, or that there is no solution to do an automated security audit.
I have been bragging for a long time that there are no organized effort (state, government, scholars or otherwise) to teach basic software security in programming. There are no ISO standard for computer security, or at least there is nothing really relevant being taught. Security certifications don't really mean anything, or it seems nobody cares enough.
I guess governments will have to get preoccupied with those things, because you can clearly see that for now, it's an arms race kind of strategy, as long as the NSA has the upper hand, things are okay, but if it becomes a bigger national security concern, things might change. I guess you can smell that the russians are getting good enough and that it's time to make internet security a more prevalent matter for developers.
- qubex 9y agoHow can you hope to statically analyse something that executes non-deterministically and whose state depends on the goings-on of all the processes simultaneously hosted on that system? The whole point the author is making is that there is a step at the bottom of all formal reasoning hierarchies that doesn’t behave deterministically and thus undermines formal reasoning about everything built above it. It’s just that we had never recognised it as existing because we conveniently abstracted away this reordering of instructions and thus had a model of the runtime environment that doesn’t match with reality.