What was once attacked with NP-hard algorithms now has a constructive one.
Reminds me a little bit of the lottery problem - one has to find the minimum sized set of tickets that guarantee a match of at least K numbers on any of the tickets. Statistics can easily get you the expected number of tickets you have to buy but finding that minimum set can be mapped to a minimum set cover (where you're trying to cover all tickets with your set so that tickets in the set match all possible tickets in at least K numbers).
Then, all of a sudden maths can put bounds and direct formulas of how many tickets your set has to have (not trivial).
Leaving you again with a constructive polynomial algorithm.