IIRC the argument was "we can't say it works because we have no relevant tests at all". That's what I remember from March, but I may be misremembering.