In that case you should invest directly in base technologies. The private sector will find the most profitable uses for them, and usually there are more profitable applications for technology than space. Everyone loves to talk about all the new technologies which came out of the U.S. space program, but imagine how much more we would have gotten had we invested the same amount of money directly into medical technology, material science, and orange-flavored powdered drink mix.

I'm with you on spinoffs argument, however we're concerned with technologies of specific usefulness in space to tap space resources. What is the profitable application of a zero gravity refinery for turning heterogeneous rocks into aluminum? Assume the process is (at the small scale) around 5% as energy efficient as electrolyzing bauxite and requires a high vacuum. Chances are such a thing could be worth something in a world without cheaper ways to get aluminum, assuming you could work around the gravity difference. Not so much in a world with abundant bauxite, gravity, and an atmosphere. So there is little incentive to develop in that direction unless you are actually planning to use it in space, where it would be highly useful (because aluminum is so useful in the service of energy collection in space that 5% energy efficiency actually wouldn't slow growth by much).

The technologies required for that are various things which are beyond our current abilities. We can't even do self replication on Earth. We may as well start with the fundamental nanoengineering and artificial intelligence domains. We don't know how space tech and missions will evolve, so if we try to make applied technology for current missions then much of the effort will be poorly targeted and less useful. It's already clear that more serious basic problems in materials science, AI and other domains must be overcome for space exploration to provide positive returns, and those are the fields which both the private sector and the government are less interested in supporting (due to long term horizons and riskiness of profits for the private sector, and lack of politically sellable 'results' for the government).

We actually do facilitate the replication of machinery, with the aid of human labor. An orbital factory wouldn't have much time delay compared to the human nervous system, so the minimal requirement for a fully self replicating space swarm seems to be telerobotics sufficiently good to mimic the human hand well enough to perform maintenance and assembly tasks. No new advances in nanoengineering or artificial intelligence are needed. However, until such a system replicates itself adequately to return results, it would be a monetary sink rather than a source of profit. It would become profitable at some point, because the cheap on site energy, ready made vacuum, zero gravity, absence of atmosphere/weather, reduction of rent/crowding issues due to 3d construction, enhanced transport logistics between factories due to vacuum/zero gravity, etc, would all contribute to making it more efficient.

The most efficient way to colonize space is with self replicating robots/factories. Human settlement is probably going to be more of an afterthought, or along the lines of on-site repair and teleoperations personnel for the robots doing the heavy lifting. The concept of slave labor in orbital mining colonies doesn't make much sense outside of science fiction.

Once a certain critical mass has been reached in terms of having systems that can convert raw space materials (asteroids, lunar regolith, and so on) to more useful configurations of their raw elements (metal, solar panels, breathable air, and so on) it will make more sense to set up habitats. These will probably be between planets, not on planets.

This will have some positive consequences and some negative where x-risk is concerned. On the negative side, access to high energy weapons is almost an inherent part of being in space, since it is relatively easy to create high velocity projectiles (gravity alone adds huge amounts of energy to any sizable object, including the sun's gravity), and sterilizing the entire planet will be more plausible than even nuclear weaponry.

On the positive side, we will have decentralized our population and have the ability to survive if the earth itself is rendered uninhabitable. We will also likely develop more easily decentralized types of manufacturing tech which will protect against supply chain disruption, thus reducing economic paths of civilizational decline.

It is also worth noting that the earth's carrying capacity can essentially be extended, guarding against overpopulation, by returning refined materials and/or energy from space, once the level of off-planet industrial growth hits a point where it becomes profitable to do so. So even from the perspective of people who want to stay on earth forever, there is incentive to develop this. You can also use it to guard against asteroid impacts, but that's a relatively minor gain compared to the rest of the picture.

Nope, the Moon has none of the resources required for sustaining a spacefaring civilization, except sunlight and water. Whatever resources you have will degrade with inefficiencies and damage. Your only hope is to just wait for however many years or millennia it takes for Earth to become habitable again and then jump back in a prepackaged spacecraft. But, as noted above, it's vastly easier to just do this in a shelter on Earth.

You are forgetting the rocks, including metals and so forth that we know to be present there (and on the asteroids, which are an even more serious target). Lunar dirt and rock is about 10% aluminum, just like earth dirt and rock is, and just like stony asteroids are. Oxygen is the most abundant element in them, followed by silicon. Iron is also present in small (but inexpensively magnetically collectible) amounts in lunar regolith due to meteorite impacts from metallic asteroids.

The problem with earth is that as long as we stay here, we tend to only develop technologies optimized for this environment -- which is small, crowded, and vulnerable. If you develop technologies for the Moon, that same approach will tend to work almost anywhere in the universe. You wouldn't stay Moon-only for long.

We do have approaches that could be used, but they aren't mature because we don't have a need, thanks to plentiful water and water-based geology. For example, we have long known that you can convert any substance to plasma by raising the temperature to 10,000k and the dissociated ions can be separated by mass charge ratio (a la mass spectrometry). Efficiency in such a system would be tricky, but isn't necessarily insoluble (might require that it be done at very large scale, for example). Energy efficiency itself is also somewhat less relevant given the abundance of sunlight.

The big issue with dragging our feet on space is more to do with astronomical waste than x-risk in my opinion. Every day we wait to build the first self replicating robotic space factory is another huge loss in terms of expected utility from economic growth. The chance of an asteroid impact probably isn't high enough to rate by comparison to the missed gains of even a fraction of 1% of the solar output translated to meaningful economic activity.

I'm not sure expanding into space necessarily (in the "all else equal" sense) reduces x-risk, since space warfare has the capacity to be pretty brutal (impact weapons, e.g.) and the increased computational resources that would be granted by having a mature self replicating space industrial capacity could lead to earlier brute forcing of AGI. It's probably important to control who has access to space for it to actually reduce x-risk (just like any other form of great power, really). You would certainly eliminate some x-risks entirely though (natural asteroid impact, virus that wipes out humanity, global warming caused by reliance on carbon based fuels, nearby supernova, etc.)