It's not necessarily a bad observation but if you look at a lot of dependent observations then you can't simply count the number of observations. I mean sure '

It's not necessarily a bad observation but if you look at a lot of dependent observations then you can't simply count the number of observations. I mean sure 'it has never decreased' is slightly stronger than just the fact that it increased over 2 independent 5 year time periods, but that's mostly because 'it has increased' is a pretty weak observation. The information it adds is that it must have increased quite a bit, which you already knew. So rather than a bad observation it just doesn't tell you much, hence my tongue in cheek observation that they're basing decisions on the basis that it has happened 'twice'.