This is 100% because NSF funders are not allowed to make "mistakes". They need every dollar of grant money to lead to published research, so they almost always bet on the established horses. Simply changing the rules to allow funders to bet on longer shots would fix much of this.