Having the equivalent jq expression in these examples might help to compare expressiveness, and it might help me see if jq could “just” use a DFA when a (sub)query admits one. grep, ripgrep, etc change algorithms based on the query and that makes the speed improvements automatic.