Sometimes your longest serving engineers can be your biggest anchor. Good engineers are (justifyably) highly opinionated about what can be done, but sometimes it turns into "what I do works, so all other ways are wrong". At that point the best move for them might be to go learn how somebody else does it. Wish them well, and back a different horse.

The hard part is "that add value". Not only it's hard to measure, but highly depends on scope of what is done

Counting commits is exactly the type of stuff starting to happen. linearb.io for example. I’m seeing this used as a way to rank dev value under the guise of “we’re just trying to help teams improve throughput”

edit: corrected url