The key to any large scale engineering seems to me to be to first create a self replicating self maintaining set of machines, then just feed them mass for 50 or so generations. If a doubling time of two weeks can be achieved (which seems quite possible), then in two years from building the first set of machines, everything changes.

The moon seems like a good source of mass — low gravity well, very little atmosphere.

Aircraft in the upper atmosphere of Venus is a much safer and more accessible alternative to interstellar or interplanetary craft — where your suggestion of any malfunction guaranteeing death is much more applicable — where you have to consider you have to transport your own oxygen, shield from cosmic radiation, heat the craft from near zero kelvin to our comfort zone, artificially generate gravity,.. you are dealing with a much more complex craft, in a much more dangerous environment — one air leak and you are in trouble — where repairs and reconstruction are MUCH more difficult, and much further from Earth to rescue if your ‘artificial ecosystem’ starts to fail.

On your consideration of Mars or the Moon as less volatile — consider how many seconds you would survive on Mars or the Moon — with no atmospheric pressure — if your eco-bubble ruptured. The Venus proposition by comparison is a solid one.

With the above given necessities, one has to ask the simple question — why go to the trouble of constructing floating crafts and putting them into the atmosphere of such a dangerous planet where any error or malfunction guarantees certain death?

If we have the option of building craft with onboard aggricultural facilities, artificial eco-systems, why not build them in gravity-free space within the solar system away from planets, with systems to forewarn again meteoric or comet activity to move the craft out of the way as and when necessary, where repairs and reconstruction as and when needed can be carried out with lesser risks and without disruptions?

Until we develop the capabilities to create artificial eco-systems on less volatile spheres like the moon or Mars or the technology to travel at hyperspeeds to reach similar planets as our own, artificial worlds in space would be the safer and more practical options. We’ve already developed space stations, its time to work on building up upon that experience. No, its not easy, and trillions of times easier to talk about it. However, what is most feasible ahead is the litmus question.