It's not a line. It's roughly a spiral. See https://www.skatelescope.org/l.... 3000 sqkm is probably the area of the radio-quiet park. But it's population, apart from the astronomers and engineers is tiny

The telescope when finished (2025) will need more total bandwidth between its antennae than the entire remainder of the internet is projected to need at that time.

It will be dedicated fibre, about 50 000 km of it.

What we're talking about here is connecting the (very few) isolated farms and villages within one or two hundred miles of an antenna. With a population that
distributed it isn't a last mile issue it's a last hundred miles issue.

The population is VERY spread out -- that's why they put the telescopes here in first place. Think rural Nevada, then take 90% of the people away.

Why stop at the moon? You could put half the array at Neptune's leading Trojan and the other half the the trailing one and synthesise a REALLY big aperture.

Seriously, the answer is cost. It's expensive enough building this many super-high-quality dishes and associated support structures and installing and operating them in empty (almost) deserts in Australia and South Africa, plus the 50 thousand kilometers of optical fibre to link them up and the multi million core supercomputer to do the aperture synthesis.

Putting all of that on the moon would cost trillions and take decades. The signal would be cleaner (except that you are outside the Van Allen belts so you have to worry about solar radiation) but the signal on Earth is good enough to do the science. Also the moon is actually too small. Even if you spread the dishes over the whole far side you couldn't get as big an aperture as they get with part of the array in Australia and part in Africa.

What they're going to get is years of work associated with the building and more years, but of less work associated with the operation, plus probably things like roads.

The area is extremely empty in the first place. That's why they chose it for (part of) the SKA.
The antennae will have dedicated fibre connections (the bandwidth needed for the aperture synthesis is, um, scary, but I suspect
running fibre or copper from there to every village and isolated farm would be stupidly expensive. Carefully chose satellite equipment will broadcast very
little outside it's beam, and on quite specific wavebands.

The article admits that it's not perfect (latency, download caps) but it's better than nothing and imposing radio quiet was an absolute condition of South Africa getting part of this very high prestige project.

It's even better than that. In exchange for a discount, most people would settle for a charging outlet that guaranteed (say) net full charge between 8pm and 6am. That is it might charge for three hours, draw for one and then charge for two, or charge at half-rate for 8 or whatever suited the grid.

More overall capacity might be needed, but this kind of thing makes it very flexible.

Really?

Hard to know where to start. Firstly the whole landing was just a small part of the mission. The orbiter is still up there and, all being well, will follow the comet in to
perihelion, observing all the way.

Secondly, think about the trade-offs of planning a space probe. You can make things more robust and more redundant, design more conservatively, etc. reducing the risk of things failing, but that costs you mass and power (and possibly money) which are rigidly limited. So you would have to take fewer instruments. The design optimises the expected science return by taking some risks.

The lander was intrinsically high risk, because no one had any idea what the surface of a comet is like. They had to gave it a bunch of different ways of hanging on designed around some plausible guesses. The lander has no propulsion at all (those mass trade-offs again), so it has to put up with wherever it hits. They knew solar power on the surface was uncertain, so they had enough juice in the non-rechargable battery to do the highest priority science.

In the event, two systems failed -- the cold gas hold down thruster and the harpoons. No one knows why yet, but building systems on a very tight mass budget that can work after 10 years in space is not easy. In addition, the surface of the comet seems to be harder than anyone really expected.

Given the challenges, getting any science at all back from the lander is amazing and a bonus to the main mission which is the orbiter.

My guess, before 40 years there'll be a spray-on solar PV coating you can put on your existing roof. Basically like a printable OLED in reverse. We shall see.

The cost is down to reasonable levels already, and a lot of it is installation, which would be much reduced if you were reroofing or building new anyway. But I was looking 10-15 years out, by which point they are on track to be so cheap they probably undercut decent slates.

HVDC (high voltage DC) transmission lines are economical up to several thousand miles. Hydrogen pipelines even further. You can bring solar North from New Mexico or hydro down from Alaska to the great plains.

You can also move demand. The Icelandic economy is an interesting one. Apart form a certain amount of dried cod, they basically have only one natural resource to export -- energy. They have geothermal and hydro coming out of their ears. However they're in the middle of the Atlantic, so they can't build power lines to anywhere with demand. So they have to find proxies for the energy. They import Bauxite and export aluminium and a few similar processes. They are looking very hard at exporting liquid hydrogen, or chemical proxies for hydrogen, like ammonia. The most fun one though is search results. People are building compute farms their and exporting the results.

New and replacement roofs could be dealt with by building codes. That gets you a lot of the way in a few decades.

In my view, saving the planet from catastrophic warming is a totally appropriate use of government power.

The cost (per GW) is coming down incredibly fast at the moment as production ramps up. Pretty soon it will be comparable to new
coal plant, even ignoring carbon costs.

You are right that not every roof is ideal. With current technology you can usefully cover about half of roof space. For new houses it's not hard to use a single-pitched roof design facing South for many buildings (or a flat roof on which you can put pitched panels.

Eventually, I suspect the solar cell becomes a layer that you paint or print onto every roof tile you make, and is probably cheap enough that you just use them everywhere and don't worry about it.

Modern designs don't have gearboxes.

There will need to be a range of solutions certainly, but there are lots of candidates. They need proving out at scale, and not all will succeed but a few examples:

Pumped water storage will hold gigawatt hours easily,
hydro plants can be designed to let you take their (fairly fixed) annual capacity out in bursts, if you like..

  Denmark is a bit flat, but it's also not far from Norway.

On a timescale of days you have some warning from the weather forecast, so you can shut down some industrial processes
and you can spin up cheap gas plants. Since the gas plant is just backup it can be relatively cheap and inefficient. If you are that purist, you can regenerate the methane from CO2 and hydrogen (from electrolysis) or make it from organic waste.

It's not simple, but the existing grid isn't simple. We need to devote the same energy, plus modern IT to solving a slightly different problem.

There are real problems, but there are also solutions. You can do much more to control demand on shortish timescales. No one will notice or care if the aircon or heating to their huge office building switches off for a few minutes, or if their electric car only charges 90% of minutes it is plugged in for.