The problem with that point of view is that it's not acceptable if a kilobyte per kilogram is not the same thing as a byte per gram. If ever that situation occurs, something has gone horribly wrong somewhere.

As a more direct example, network transfer speeds are customarily measured in real SI units; your 100 Mbit/s network card can transfer 10^8 bits per second, not 100*2^20 bits. Now how long does it take to transfer a 1GB file over a 100 Mbit/s connection? The answer had better come out as 80 seconds. If it instead comes out as 85.9 seconds because the 100 Mbit/s meant 10^8 bits per second and the 1GB meant 2^30 bytes, I think we can all agree that we're doing SOMETHING terribly wrong.

How does that work, exactly? Not following the citation standards is not plagiarism. As long as a person denotes in any way whatsoever that a given thought / phrasing is not his own, it is not plagiarism, no matter how wrong the form or how limited the information in this declaration. It doesn't matter if the title of the original book, or the publisher, or hell even the author's name is missing -- as long as one writes "this is not mine" it is not plagiarism. It may not meet the requirements set by the course and therefore warrant a failing grade, but academic honesty violations are out of the question.

This. Didn't you hear? It's okay to watch My Little Pony nowadays.

Again, prove it. If reducing from the undecidability of the halting problem, of even the incompleteness theorem, show how.

No, he proved that you can't decide whether a program works correctly. It does not follow that you can't prove correctness for particular programs.

Please explain how. From the undecidability of the halting problem you can derive that there can be no program that reliably decides whether a given program is secure, but this is a very long way from saying that it's impossible to prevent all security holes.

What keeps me from trying to prove a given program's security by hand? Or how about a program that can prove security IF the input program if sufficiently well-written, giving up on secure but unclear programs for which it can't be sure? Turing's argument doesn't disallow that, and in fact it's perfectly possible.

Oh? Please prove it, then.

No, it assumes memory for which all access are *at worst* a certain equal cost - in other words, memory for which accesses have an upper bound. Knuth's analysis still holds if certain pieces of memory are FASTER than the norm. If we take memory access going through swap as the normal, worst case, upper-bound cost of memory access, and RAM and cache hits as special better cases, the analysis still applies to the real world.

Doesn't it? The theoretical worst case corresponds nicely to the real-world worst case of all memory coming from swap. The (typical) case in which there is also a bunch of fast RAM and even faster CPU caches is just not the worst case.

Really? It is exactly the reason that made me never buy a console: consoles have all the DRM nastyness PC games have, except that I can't avoid it with cracks.

It is entirely accurate - assuming you have access rights to the target process. To summarize it mostly accurately, you have access rights to the target process if it's yours (started from your account), if you have admin rights, or if you have global debugging rights (which requires admin rights to grant).

In other words, it isn't insecure at all. Of course, this point becomes moot if the malware runs from the same account as the user, or even with admin rights, as is common on Windows. But that's an entirely different problem which is orthogonal to the issue described here.

Note that the truth of your statement does not change when you replace "R" by "python" (and remove the word "statistical"). Nevertheless, I would still call python a programming language.