FWIW, #5 is the only one I hadn't seen someone suggest as an application for blockchain before I read this.

The first is how every language starts out. The second and third are the same thing and have been the goal since the beginning.

In unmanaged languages, it's generally easier to know/control when allocation happens

They are everything but a niche.

Their products may have a lot of users, but I can't say I've seen all that many job postings.

Not really, as memory allocation is performance wise very cheap :)

Unless your VM can't allocate things to two parallel tasks simultaneously, causing allocation to effectively sequentialize your code.

I defy you to write a compiler or kernel without such knowledge.

I had such knowledge when I first wrote a compiler, but I don't recall actually using it anywhere. What did I miss?

Getting a denotational semantics right takes a lot more than familiarity with programming in the kind of language you're specifying. You might be better served by reading up on order theory and logic (get a text that's really pedantic about everything).

Looking over the household repairs/replacements we've had to do the past couple years, I can't find anything a 3D printer would have made easier. Faucet aerator has its threads worn down? A new aerator is dirt cheap. Leaking at the base of the faucet? Spend a couple minutes with a wrench. Thermostat's temperature sensor is dying? Not going to be printing one of those. Wallpaper peels off? That's a job for a rather specialized 2D printer, or in our case, a can of paint.

You might want to give this a read.

So tech companies don't want to be in high crime locations in the middle of neighborhoods that most of their workers wouldn't want to live or send their kids to school?

TIL Roxbury is the entirety of downtown Boston.

I do remember roughly what a language can do

Then it seems you do remember something about the language's semantics. Maybe it's the details of syntax you're forgetting?

I was willing to give it that interpretation too, but then I got to the part where Bates suggested that the court should choose the adversarial advocate.

It doesn't really take much for that to be the case. Remember, there is no way to decide whether a Turing machine with a given input will reach a given state. All you need is for there to be _some_ undefined behavior in the dynamic semantics -- "division by zero is undefined" is sufficient.

I have never seen a compiler that does that, and I seriously doubt if is really common.

I'm a bit depressed to find a /.er who's never seen GCC :-P
I once wrote an overflow check wrong -- I tried to write an `if' that would check whether the preceding operation on signed integers had overflowed. Overflow on signed integers is undefined behavior, so once it happens, it is legal for the program to do anything. "Anything" includes updating the variable with the overflowed value and then skipping the condition check, which is what GCC's output code did.

The jury heard the testimony from all the witnesses. They saw and heard all the evidence. THEN they wanted to punish Toyota. Yes? So what's wrong with that?

The jury's function is not to mete out punishment. It is to determine whether the defendant committed the wrongdoing they are accused of. When a juror expresses desire to punish, it makes me wonder whether the verdict was motivated by epistemologically sound consideration or by that desire.