[squibbles]

I have seen many people fail to realize (and sometimes to understand at all) that taking measurements of large numbers of people is always fraught with errors of all sorts. No matter how hard you try, it is like the universe always pushes back to make sure the data are imperfect. Sometimes those imperfections are acceptably small, and other times the imperfections are uncomfortably large.

In my opinion, anyone involved in such large-scale data collection and analysis should acknowledge the inevitability of error and provide disclaimers about potential sources of error.

[fab1an]

Yes, but if the error consistently results in you getting paid more than you should have, it's not really an error, is it? No, we would in fact call that fraud.

[hari_seldon_]

+1, the issue here is not that errors can occur but instead how a company responds. It’s tough to be sympathetic to leaders in a company that were aware of the errors and then took no action for years to remedy it, particularly as the error resulted in overcharging customers.

[Analemma_]

Funny how these mistakes never seem to undercount ad impression numbers though.