Success rates for D and F templates. (A) We see the proportions of successful intervention attempts for D-based templates of four sizes across six different network sizes. (B) We see the same for F-based templates. Both subfigures include the performance of the single-variable approaches for comparison. While D-based templates perform reasonably well in the smallest network sizes, it is clear that only the F-based templates show consistent and significant performance across all network sizes. Note that 4-variable templates are not included on the 40-variable network for computational reasons.