What percentage of automated acceptance tests have bugs, so that the first failure signifies a problem not in the code, but in the test?

July 20, 2009 from EventBox