In practice I think it comes down to whether you're trying to minimize risk or maximize reward. The former, conservative, strategy values past experience, since a known entity would seem to reduce the possibility of massive failure. The liberal strategy values future potential, thus attempting to optimize on likelihood of massive success, but simultaneously increasing the chance of massive failure.
If you want to make X, think of it as a machine, isolate the core parts and tools. Then hire candidates based on past deliverables and years experience with what you think's needed to make X.
If you want to disrupt X, hire people with a) ideas you subjectively find interesting, and b) valuable skills and knowledge which exceed your own with those tools/topics.
In my experience, as hiring groups grow, they become more conservative. What a company looks for in Engineer #1 is different than Engineer #1000. The more opinions involved, the more the group seems to decide based on risk rather than reward.