To me the tests have to be real measurements rather than a proxy. Having a special test makes the test easy to pass. If the test was instead to have sampled d

To me the tests have to be real measurements rather than a proxy. Having a special test makes the test easy to pass. If the test was instead to have sampled data from x,000 cars, gathered at service intervals fit within certain bands, then that would require actually engineering most cars to return within the set range across a wide range of use types. Smaller niche manufacturers could just pass a simpler test to encourage market entrants. 200 slightly worse cars is nothing against 200,000 intentionally bad cars. The regulations need to target what is trying to be achieved rather than simplified proxies.