I thought the point of the Acid tests weren't to reflect reality but to encourage web developers to support useful but unsupported parts of the standards.
That seems like valid reasoning to me. I've also seen Hixie make comments that seem to indicate him wanting to avoid these issues for any potential Acid4:
When we do Acid4 (probably around the time we have at least three major browsers shipping Acid3-passing browsers), I think we'll have to focus on testing fewer, more critical things. Acid3 tests a lot of critical stuff, but also checks a lot of less important stuff at the same time, and it's in those areas that we've had the most problems with specs changing under us.