Active Internet topology mapping depends on carefully selecting traceroute targets, since exhaustive probing is costly in IPv4 and infeasible in IPv6. This work implements and evaluates two target-selection strategies: a classical baseline that probes one representative per IPv4 /24 prefix, and an adaptive hierarchical strategy inspired by Subnet-Centric Probing. The hierarchical method recursively partitions BGP prefixes and continues probing only subprefixes that reveal AS-level diversity. For IPv6, the method is adapted with hitlist-based representatives, priority-queue exploration, and explicit probing budgets to handle address-space sparsity. Experiments using BGP inputs ranging from 10,000 to 200,000 prefixes show that the adaptive strategy substantially reduces measurement effort in both protocols. In IPv4, the 200k-prefix experiment reduced the probing load from 8.1 million to 299.7 thousand probes and the execution time from more than 60 hours to about 9.5 hours, while still observing 1,081 of the 1,216 ASes found by the classical method. In IPv6, the largest experiment reduced more than 4 million probes to 239 thousand and execution time from nearly 29 hours to 44 minutes. These gains, however, come with reduced prefix-level observability: audits show omitted IPv6 /48 blocks and specific IPv4 outliers where broad prefixes were pruned too early. The results therefore characterize hierarchical probing not as a complete replacement for exhaustive baselines, but as a scalable cost-coverage trade-off for AS-level topology discovery.