The process has been discussed at length on the forum and on discord and is currently the one in place.
I 100% agree that there is merit in the verifier approach, while raising the point that it is probably not universal (some techniques may not have any suitable verifier).
My opinion is to move to this more agile governance model (i.e. validating machines on the fly based on verifier data) once we reach a small set of undecided machines, which, given the projected results of current methods that are discussed on discord should be around 10,000 machines.
Finally, note that the restrictiveness and constraints in the current process are thought to ensure maximal trust in validated techniques and do the ground work necessary for publication (reproductions, formal proofs, etc…).
You are more than welcome to express your opinion in the thread and/or discord.