Talking scaling, 0.4% on a sample of 50 is pretty much nothing.  However there’s a reason some companies try and maintain a six-sigma process control that considers even 3 defective parts per million opportunities a minimum. Something that NASA goes above and beyond in their stuff.  Once you’re handling something that scales up, that “oh that’s not so bad” starts to become a big number.

Looking at one end for nice numbers can look good, but until you’re looking at the other end of things and side-effects which perhaps may not be so numberly which is where things can happen down the track can get predator-face ugly.  If it was a simple as pulling a test and taking some numbers, we’d be building tanks out of glass, and have chefs being replaced by robots.

That’s why the people in the know when they’re asked a question already know that something’s a pretty bad idea in the first place, before the somewhat small numbers begin seeming insignificant.

What does a number mean when the people making the decisions don’t understand or comprehend the weight on a value?

What’s the chances that this Internet filtering thing in Australia has been brought so far based upon the misunderstanding of a few ‘small’ numbers?