Erm... that's like arguing that the bus is dead because people are investing in taxis. Turbine systems of this sort are expensive and inefficient, but they're a necessity because of supply and demand patterns. Utilities are "shitting themselves" for a least-worst option here, as the current load issues put physical strain on the whole infrastructure. Making the supply side of the equation less predictable only makes this more of a problem.

Yes, but you still need to have variable, reactive supply to deal with that 50% variance. Reactive supply is less efficient than constant supply, so there's no net gain....

Thousands of electrical fires say you're wrong.

You've got most of the right answers in there, but a little mixed up. The aluminium foil survives because it has low resistance, so little power is delivered over it. Electrical power is distributed across a circuit in proportion to resistance : P = I^2 * R. Low-resistance components (eg wires, and your foil is just a flat wire) are therefore less susceptible to oversupply. But put more power into the system, and more power will be delivered to all components, and some of them will fail and catch fire. Don't believe me? Get a 3V filament bulb and connect a 12V battery to it -- it will pop because too much power is delivered across the resistive filament.

Ridiculous? Not at all. This was a known issue when my parents were getting an extension built in the 80s and wanted a wind turbine. It wasn't possible to connect it to the grid at all back then due to the risk of substation fires, and the technology available wouldn't have powered anything useful, so they didn't get one. Since then, a lot of money has gone into improving the grid here and developing smart controllers to prevent problems, but many parts of the world still have networks that are physically incapable of dealing with micro-generation.

Stopping people all trying to flee down the same road can keep it open to emergency services, saving lives as a knock-on effect. Furthermore, jammed roads are usually the result of panic, and the first priority after a sudden disaster is to avoid panic. It doesn't matter that your phone may be flat in a few days -- the first few hours are the crucial point. Prevent panic by giving basic information and reassurance; telling people where to go and what to do.

I'd point to the London tube/bus bombings, myself. No damage was caused to the networks, but the number of people phoning here, there and everywhere overloaded the system. Emergency services' bandwidth was successfully preserved, but lots of non-emergency calls couldn't get through. Being able to tune in to a local station and find out what the hell is going on in such circumstances would be a reassurance, if nothing else.

Norway is unusual in that people actually use DAB sets.

Even dumb-phones have FM in their radio chipsets.

It's not just about phones not working, but about capacity. If there's a big enough emergency (eg the London tube/bus bombings) the number of people using their devices will max out the network, even if every single transceiver is still up and running. Many people's first reaction in such situations is to stand for half an hour on redial trying to get a phone call out. FM would mean being able to "broadcast" information to a lot of people at once, and discourage them from clogging up the network.

There are some scientists who think that the idea of microbes "developing" resistance is wrong anyway. All our antibiotics are naturally occurring, and a mouthful of soil contains dozens of different antibiotics. The alternative theory is that all we've done is change the balance of antibiotics in the environment, leading to an unnatural selection of antibiotic-resistant strains. The abundance of life in the rainforest extends to antibiotic-producing fungi, so the microfauna will naturally have been exposed to a broad variety of antibiotics, and therefore natural selection will have led to resistant strains.

From TFA:

Computer Vision research has been behind other AI areas in its use of generalisable code, but current AI and machine learning algorithms are effective and efficient enough that image processing with them is finally practical. That's the novelty.

Even if you wrote wrappers to access all their libraries in (for example) Python, operating an inference engine and manipulating the results returned would be a complete pain. This is a high-level language designed as a problem-solver over a probabilistic domain. That in itself is a very far-removed abstraction from the processor's opcodes...

Your problem is that you're thinking to heavily in terms of C-like imperative programming. Picture sounds more comparable to Prolog to me -- an automated problem-solver. The core of Picture is an "inference engine" -- the core of Prolog is an inference engine too. The difference is that Prolog operates in pure predicate logic, whether Picture is a probabilistic language.

Probabilistic programming in even a relatively high-level C-like (eg Python) would be a total pain, because you have to explicitly call all calculations. An inference engine will do the calculations implicitly and will prune any unsuccessful branches off your search tree without you having to worry about it. The end result is very complete solutions to problems with a low bug count, but (in general) relatively high execution times.

This sort of very-high-level language isn't suitable for every task, but it serves its task exceptionally well.

Er, OK.