I always wondered though, the way we correlate numbers to driving ability seems sketchy. Some people take multiple large dabs of cannabis concentrate - and they're barely feeling it. This is because they basically do this all day every day.
Someone like that would fail this test, despite their motor skills remaining near baseline.
It's clearly different for alcohol, because it directly increases "network latency" (if you will) of motor control and sensory inputs.
So ultimately I would like a push to improve roadside driving ability testing, regardless of suspected substance use. This covers sleepiness, dehydration, etc.