Most drunk driving fatalities occur at BAC levels way above the 0.08 limit (I don't have the source handy, but I think 50th %l was over 0.15), even though comparatively few drinking drivers have such high BACs. Any resources spent going after drivers with a 0.08-0.09 BAC are resources not spent going after drivers with a 0.15+ BAC, even though catching the latter drivers would do far more to improve public safety.
I did a research report as a senior in high school (1982) about BAC levels. The ABC (Alcohol Beverage Control board) sponsored the contest and I placed high enough to win $300 in scholarshop money for college. Anyway, all my research was done by interviewing people. One of those was Dr. Patricia Waller who was head of the Highway Safety Research Center at UNC. They did a BAC test using med school students who, as it turned out, rarely drank. She said virtually all passed out by .10. Dr. Waller then had an idea. The building was being renovated so she asked some of the construction workers to volunteer. She chose those who drank regularly. At .10, she said these people barely showed any impairment. I also spoke with the head of the NC Highway Patrol and even he conceded that BAC levels were not a good measurement of impairment and added that it was easier to convict with a quantitave benchmark.