I'd argue that you need a loop that's capable of operating without Earth support for a significant length of time - vastly beyond what they managed with Biosphe

I'd argue that you need a loop that's capable of operating without Earth support for a significant length of time - vastly beyond what they managed with Biosphere 2 - for margin of safety reasons. Possibly also with a very different, scaled down design - maybe not growing all their own food, for example - because 13,000 square meters is maybe a bit much for a first off-planet base. And the best place to start working out how to do that is right here on Earth. Because it's an immediate cost reduction of at least one and probably several orders of magnitude, and because you can much more easily iterate on the design while you're still working out the kinks. And then, once you've got it working well here on Earth, then you have a go at doing it on the moon. This idea that the best place to get started on working out how to do these sorts of things is on the moon makes about as much sense as suggesting that the Space Shuttle program should have had astronauts rushing to work out the kinks of EVA missions in space instead of doing everything they could to test and practice procedures in swimming pools in an effort to learn as much as possible before you blow half a billion dollars on trying it out in orbit.