> The difference is that the colony needs to be able to survive indefinitely,
> so that it can serve as the starting point for a new civilization.
So, more precisely, it must be able to support humans until they can upload or otherwise become less vulnerable. This is far from indefinitely. But, if a colony can suport itself for several decades, I think it's safe to assume that it can also support itself for much longer periods of time (or "indefinitely"). Especially if you start terraforming Mars (perhaps using nanotech or comets etc).
> That
> means it can't get by on a stockpile of spare parts. Ultimately, it must be
> able to build all of the equipment it needs to survive, and all of the tools
> needed to make that equipment, and the tools to make those tools, ad
> infinitum.
So what you need is a working (nano) replicator, some basic assembly programs (for the most important replacement parts), programming equipment and some flexible, all-round experts to design new programs. Perhaps it would even be feasible without nanotech, but as fairly advanced (see below) nanotech would most likely already exist, this can be ignored for now.
> What exactly do you mean by "functioning nanotech"? I ask, because Ithink
> your assumptions about what you will be able to do with it are very
> different from mine.
>
> What we are implicitly assuming here is a future in which AI is a harder
> problem than nanotech, and therefore we don't have to worry about the SI
> issue until after we have survived the gray goo threat. In that scenario,
> early nanotech is going to be rather limited in what it can do.
If we assume that gray goo (or some more other form like black goo, which is IMO a more serious problem) is the problem which ultimately forces us to migrate, it follows that we have at least goo-level nanotech. This means fairly advanced, autonomous replicators which are sturdy enough to function in the "real world", not just in vitro. These nanites (perhaps minus the self-replicating feature, for safety reasons) should as least be capable of mining Mars for various materials and water, construction of simple structures and life support functions like oxygen production (perhaps even help to terraform the planet). Computer-controlled in vitro nanites could presumably do more complicated tasks like fabrication of machine parts or whole new machines, for example. Various useful programs (for the replicators) could no doubt be brought from earth, so that you wouldn't be starting from scratch.
> Early assembler technology may be theoretically capable of making anything,
> but in reality it would require a lot of human effort to make it work.
> First you have to actually design the things you want, in some kind of CAD
> program.
Some 20-30 years from now CAD programs should be pretty good.
> Then you have to figure out how to assemble them out of whatever
> molecular building blocks your system uses - hopefully this process will be
> highly automated, but it will still have manual steps if you want to make
> anything complex. You may be able to make a new valve by hitting a button,
> but if you want a new kind of robot its going to take a lot of work to
> actually make it.
Yes, but even if we assume that this will take a lot of human work, it is typically something which can be done by small teams of experts or even individuals within acceptable time limits (months perhaps). A Mars base is as good a place to design replicator programs as any.
> Also, nanotech only addresses the manufacturing issue. We still need to
> locate resources, mine them, maintain our equipment, and so on.
I think the goo-level nanites could handle this just fine, certainly with the help of tele-operated vehicles and human intelligence to deploy then in the most promising areas.
> Nanotech would greatly simplify all of the manufacturing problems, but
> without human-equivalent AI it doesn't help much with all of these other
> issues. So, our early-nanotech colony might only need a few tens of
> thousands of engineering types, but we still have to have the whole gaggle
> of scientists and assorted other workers.
Ok, say some 100.000 - 200.000 people? Let's assume that this were the case, perhaps a viable setup might be to start asap with a small, (mostly automated) commercial Moon base, use that to mine materials and produce components for a "space wheel (-esque)" colony capable of supporting the above mentioned number of people. The wheel could then set out for Mars where it would be put into a stable orbit, while some of its inhabitants would go down and start mostly self-supporting settlements. Could this approach deliver results within approx. 30 years on a "realistic" budget?
> As an interesting note, I will point out that no one has ever even come
> close to building a real, completely closed, indefinite duration life
> support system. Recycling air and water has been done
Yes, and you don't even need a 100% closed system, as you can obtain your oxygen etc. locally (i.e. a hybrid system). That's why the failure of Biosphere isn't a big deal: you *can* compensate for the (initial) ineffinciencies of biological life support (most likely algae, some 8 square ft per person to reach equilibrum, if I remember correctly. These are old NASA results, we could probably do better than that). If Biosphere has tought as anything, it is that one should be minimalistic; use only the most efficient and easy to maintain plants, and don't try to create "natural ecosystems". Not for life support, anyway. Also there's obviously the "concrete" issue and the fact that you must always have means to supply the system externally, up to 100% for at least several weeks.
, but no one even
> tries to tackle the food production problem. The most plausible approach to
> that for a Mars colony would be to build greenhouses and/or hydroponics
> facilities, but that requires shipping a substantial amount of equipment
> from Earth (which is why I expect the life support mass requirements to be
> so large).
Greenhouse components should be fairly easy to make with even primitive (nano)replicators, or a fairly limited amount of today's industrial machines (and a considerable number of prefabs could be brought along). Neither this nor life support in general are serious problems, IMO. The really important issue is whether a relatively small, isolated group can bootstrap itself towards a Singularity. With some form of moderately advanced nanotech, I think it can.