> I recommend going and collecting some data, Mr. Epistemologist.

My overriding point is that “collecting some data” in this industry is fundamentally faulty.

It’s not that we have zero anecdotes or data. It’s that we can’t know if the little data we have is truly representative of the wider data set.

> You probably know more of those people than you think...

Of course that is likely. That just reinforces my point that there is more that we _don’t_ know about statistics in the industry than we _do_ know.

> That's >x10 right off the bat out of the people I actually know. So that's the bottom estimate.

You haven’t proven that your sample is random, so you have no basis for calling this the floor.

> Is x100 high? Maybe, but not off by an order of magnitude.

How do you know it’s not off by a magnitude?

My whole point with epistemology is that this entire data set is shrouded. We can’t know if our estimates are accurate or not.

Your methodology is clearly not accurate. I don’t have any confidence that mine would be any better. But at least I’m intellectually honest enough to know what I don’t know.