Almost anything can be done if you don't care what it costs. What I don't see here (or in the similar Apple and Google announcements) is any indication of what their cost target is. Does anyone have any idea what their electricity costs need to be (or what the average datacenter revenue per megaWatt is)?

I have more than that in my basement. Think I'll get some solar cells and put out a press release.

They still do film processing (most films are still shot on film), and provide a significant amount of expertise and lab work in film restoration and archiving.

If you're asking what they've invented in the last 20 years, the answer is essentially nothing.

The law is whatever The Decider says it is.

Which is exactly what we have now.

Ah, the legal wisdom of Antonin Scalia. The man isn't fit to judge traffic court.

Analytic solutions are not necessarily easier to calculate.

Analytic solutions tend to involve special functions for which the computer can only compute an approximation anyway. Have you ever tried to write code to evaluate the error function over the entire domain of floating point numbers? (Yes, I know, it's now in the standard library; ten years ago, it wasn't.) That's one of the easier ones.

Even if there are no special functions, analytic solutions are still often harder to calculate if the problem is big enough. Think of solving systems of linear equations, one of the standard workhorses of numeric programming. We're talking really big ones; hundreds of thousands of equations in hundreds of thousands of unknowns or bigger. In the real world, this problem would almost certainly be solved using successive approximations, even though high school students know how to solve them analytically.

Finally, and most importantly, the problem statement is usually an approximation. Take the OP as an example. What this kid almost certainly solved was an analytic solution to the problem of a particle in a gravitational field with linear air resistance. Well, air resistance is not linear. At low velocities, and for projectiles with a sufficiently small cross-section, it's close enough. But it's still an approximation.

The advantages of analytic solutions are almost always not computational. What they buy you is understanding. The methods of obtaining the solution, and the form of the final equations, often reveal some deep insights about the problem. For many situations, that's far more valuable. And it's certainly something that no computer can give you.

I love ISABELLE myself, but hey, whatever floats your sequent.

I'm guessing you don't write high-reliability software.

There are plenty of situations where you can recover from a SEH situation. Division by zero is one of them; think of a user-scriptable application with an embedded interpreter. It's much more efficient for the CPU to catch the exception than for you to keep check the divisor.

Even in the case of an access violation, there are some applications where dropping the connection which caused the problem but keeping the others going is an acceptable solution.

IMO everyone should be allowed to puplish in the language of their choice, so long as they do it in the privacy of their own home.