But as nobody has suggested a criminal offence of which the contractor may be guilty, it hardly seems relevant.

Of course there's a contract: there's one described in the summary above. The contractor agreed to wipe the machines in exchange for getting them for free. There, that's a contract. Now, it may be difficult to sue him for breach on the basis that there doesn't appear to have been a *written* contract, but that's an entirely different matter from there being no contract at all.

Still: I've gone through 3 consumer-grade routers over the last 10 years, and each time I've got a new one it's because the old one isn't up to the job, not because it's failed. They shouldn't need active cooling (they don't use more than about 2W in typical use), and the power supplies seem perfectly adequate for the task to me.

Is there a documentary I could watch that will give me some idea of the absurd disaster scenarios somebody has invented for this technology?

Sure, but this article contains data showing that the difference in profit margin between Apple and Samsung phones is minimal these days, so you're clearly just making this shit up.

Switching between client and server is a large enough switch anyway. I'd say it's actually *easier* to work in different languages on each side, because it allows you to keep your conceptualisations of each side more distinct in your mind. Sure, I can't justify that in any way, but I doubt you can justify any assertion of a cost in a "mental context switch", either.

That's because there are certain ways in which Java *is* slow:

* AWT is a heap of shit, very badly designed, and takes a positive age to initialize. Swing is no better, because Swing is built as a layer on top of AWT.
* Java's performance for floating point leaves a lot to be desired because Java's specification of some mathematical operation means that processor-specific instructions cannot be used, e.g., to calculate the result of Math.exp() on x86 processors, so it must be done in software. This bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7174532 suggests they may have recently found a solution, but I don't think this is in any released version of Java yet.
* Garbage collection is much slower than hand-optimized memory allocation, and heap allocation is much slower than stack allocation, so a well written C++ application can always outperform Java on memory handling.

This doesn't mean that Java isn't perfectly suitable for the vast majority of tasks, particularly those that are I/O bound like a typical server application, but there are things that it really isn't good enough for.

(Signed, somebody who once attempted to write a high-performance neural network simulator in a language that can't do exp() at a reasonable speed.)

Not even then, I would imagine. V8 is not marketed as an enterprise-class product. There is no support for V8; it's a free product, released only because they wanted a world-class javascript engine to include in Chrome. Any other uses are practically irrelevant as far as Google are concerned.

If I hadn't already posted, you'd be getting a +1 Insightful. Well said.

Why?

What makes it more important to give up our liberties to prevent terrorism than to give up our liberties to prevent accidental death caused by wild animals? Analyse the logic behind the sentiment, and I think you will find it an utterly irrational and thus indefensible emotional reaction in a situation where a cool, analytical reaction is more appropriate (because what we stand to lose is very valuable).

The point remains, though, that going by current most-widely accepted theories of linguistics as an academic subject, there is no such thing as a universal "correct" version of English. Grammars and dictionaries are descriptions of the way people speak, not prescriptions for how they should speak. Identifying one variant of language as correct and another as incorrect is a cultural bias, and not an appropriate thing to do while studying the language in question from an academic perspective. Therefore, to make any definitive statement about English grammar, particularly one as controversial as "double negatives always form a positive", is clearly not the kind of thing you would expect an academic in the field of linguistics to say.[1]

This does not mean, however, that they wouldn't prescribe a particular dialect and set of grammatical rules that they expect their students to use. To require a particular variant of a language in a particular context is a matter of practicality, and is somewhat orthogonal to the study of the language from an academic perspective. (Consider that the students could be studying French and writing their essays in English, for example.)

[1] All of which kind-of detracts from the joke. Change the MIT professor to a high-school English teacher and it works much better.

Creative != using the latest buzzword-compliant language/framework.

The older workers just realize that switching to a new framework will usually end up wasting time that could be spent actually coming up with some interesting ideas for novel features that customers might actually care about. The productivity gains of new environments are marginal if you spend most of your time learning how to use it effectively. You're not going to hit your 10,000 hours to master a skill (per Gladwell's suggestion) if you switch to a new one every 5 years. Or at the very least, you're not going to have chance to do very much with that knowledge once you've obtained it.

While this is technically true, the effect is small for any realistic sample size and population size, as it proportional to the square root of the fraction of the population who are not included in the sample, so you might as well assume the population you are sampling is infinite (which is how these surveys tend to work out their error margins). While this may make for a sizeable difference between the reliability, say, of a 10%-of-population and a 20%-of-population sample, here we're talking about some numbers that are very close to zero.

For 1,000 people in 300,000,000, sampling population size correction factor of sqrt(299,999/300,000) results in a decrease of error of 0.0002% from those calculated based on infinite population.
For 1,000 people in 16,000,000, the decrease in error is instead based on a factor of sqrt(15,999/16,000), so is 0.004%, meaning that a random survey of 1,000 people in the USA is about 0.0038% less reliable than a similar survey of 1,000 people in a much smaller country.

Yes, but the government of the UK is continuous with the previous government of England & Wales, and the government of England before that.

As I understand it, although obviously the office-holders were changed, the Normans basically retained the old Anglo-saxon system of government, because they felt it was better than their own. Therefore, as far as I see it, this only differs from what happens in a republic when the governing party loses the election by the means of chosing who gets the power, which seems irrelevant from the point of view of determining whether the country is the same country or not. If the Normans had come in and imposed their own system of government, making England part of a Norman empire, I'd agree with you, but that's not what they did.