People have been claiming this (at least the on-line part) for a long time, though. I seem to remember software company executives in the 90's drooling over the thought that you'd pay them a monthly fee to access their word processor and photo editor apps from your thin client at home.

...what kind of hell-spawned interface did the other guy make if Facebook is clean by comparison?

In other news, my barber tells me I need a haircut.

You mean actual, non-management, I-write-software-all-day people making near $250k per year at M$ and/or Amazon? I've worked at one of those companies, and I seriously doubt it--I would be really surprised if more than 1% of software engineers make more than $150k.

Now if they could just do the same thing for Windows itself...

On the slim hope that someone actually wants to suggest this, please make sure you also include the names of any cities where they pay engineers $200k, because I want to move there.

You make it sound like the act of sneaking ~4GB out of a "secure" facility would be hard. I'm sure you could find (or easily make yourself) a waterproof capsule that would hold a 4GB SD card and/or usb stick, and stow your data device in a bodily orifice for the transit in and out of the facility. The hardest part would probably be actually finding a way to copy it off the secure system without getting caught, unless you're the IT admin guy that's trusted enough to be alone with the backup tapes.

I would think grabbing a big pile of stuff at once would be less noticeable than a lot of little collection events, too, unless there's no chance at all of getting caught while you're actually copying data.

Sorry if it appears that I'm conflating the two; I didn't mean to imply that accuracy and complexity are necessarily related. I meant that sometimes one can find ways to reduce the complexity of a correct implementation without affecting accuracy, and sometimes one will find that a numerical method has been implemented incorrectly, or that the selected method isn't applicable to the problem at hand, and therefore the results produced may be wrong.

The general case I'm talking about for reducing complexity is when the original coder chose some sub-optimal way to implement something. For example, some step in the algorithm requires using the value of an integral inside a loop, and so they make O(n) calls to an O(n) function to calculate the integral, rather than computing the integral first and then using the computed values in the loop.*

* And, no, we're not talking about small values of n, or limited memory, or any other reason somebody might have for doing this. It's just that they didn't realize they were doing something in an O(n^2) fashion when it could be done just as accurately in O(n).

I have to second this. There are some fields in which the researchers are not strongly mathematically-minded, let alone numerically-minded, and it does indeed lead to inefficient code that produces results which may be incorrect in ways they never thought to check. You could easily be a hero (or villian) if you can get into one of these research communities with good software and numerical skills: you can sometimes make their code run in O(n) or O(n log n) time instead of O(n^2) (or worse), but you may also show them that it's producing results (which they've been relying on) that are incorrect in some (perhaps not so subtle) ways.

Just based on things I've seen and read here and there, he was "much better" back then because he was confined by limited resources, and by people around him that would actually say, "No, George, just....no."

FTFY. And, of course, I know this. It's just amusing (and encouraging) to see Microsoft whinging that they're having a hard time indoctrinating students into dependence on their tools.

Boo-fucking-hoo.

That fantasy world isn't nearly as nice as my real one, where my wife enjoys pointing out noteworthy asses and breasteses for my enjoyment.

Thanks for mentioning that; I'd never run across it before. Yay for learning something cool before I finished my first cup of coffee. :)

You mean like regular light switches? Is that tunnelling taking place through oxide layers on the contacts or something?