[shostack]

Exactly. What is also missing from the post on this test is any sense of if/how they plan on addressing this piece of the puzzle as it seems a rather large one. When your experiment group is small, it simply won't cause a big enough impact at a macro level to move the inflation needle, but that is what many predict would happen.

How can we measure that or get any sense of how realistic that outcome is before a large scale program goes into effect? Don't get me wrong, you have to start somewhere, so I commend this effort, but I have yet to see a good proposal for how to test and deal with this.