"Or the government could offer to purchase thousands of tons of water in low Earth orbit at fifty bucks a pound—far cheaper than it costs today to get a pound of water into orbit—that could be used for rocket propellant and life support."

I think it would be more realistic to buy hundredss of tons at $500 per lbs.
(I love saying that:).

As I suggested many moons ago, I think NASA should offer to buy 100 tons of water which is placed in LEO in a orbit which will not burn up in the atmosphere within 5 years. Then, if and when there is enough water in orbit, NASA should offer to give it to any company that will to spend it's own money to develop a system which will convert the water into rocket fuel. [Say ownership is transferred once 1 ton of water is split and stored and/or used.]
Alternatively or included, NASA could buy water which is delivered to EML-1*. Because this requires more rocket power per lb delivered, NASA would pay more say $1000 per lb delivered for a total of 100 tons.
100 tons at $1000 per lb costs 200 million. So for as little as 1/4 billion, NASA could open up the space frontier- 1/2 billion if it *really* wants to be "aggressive". In other words less than the cost of one Shuttle launch.

*It doesn't have to be EML-1. It could be EML-2 or for a more stable orbit it could be EML 4 or 5. It doesn't really matter they are at low delta-v to one another- it could even be at GEO which is harder to lift to, and a slighter bigger delta-v to the L-points, but maybe it would fit somone's business plan better.