> Beyond that if it sparked discussion that either spurred the uncovering of real data or identified the requirement of further considerations, it's still of value.Which it didn't. We're not going to get better data until Google releases it, and Google is clearly not going to be doing so.
It most certainly did spark discussion – I'd point out, somewhat ironically that you've been involved in one of them yourself – but there were a number of other replies. Some of which (mine included) addressed the data issue head-on.
> But, more fully, anecdotes aren't speculation.
This claim would be okay if the anecdote was offered in a neutral environment.
There is no such thing as a neutral environment, so your premise fails from the start.
And, no, hits to a web page or unique visitor counts are not tantamount to real usage figures for a social media site. So the analogy to a TIOBE survey breaks down as well.
I disagree that they are useful here.
Well...
I said, explicitly that I was addressing the larger point of the other poster asking "Why is there the need to respond to data with anecdotes?" I started directly by pointing out the failing of "data" in this case – I find it to be an over-played trope to always say "anecdote BAD!" and yet people fall all over themselves if someone puts a shitty graph or table up, not stoping to consider what the "data" means and what possible failings it might have.
And that led me directly to the value of anecdotes in general terms – something you now say you agree with, so we'll call that point made.
Given that we agree about the general value, I think it's clear that in the face of crappy, blindly accepted data, a mere (and admittedly weak) piece of anecdotal observation does, in fact, point out that the emperor has no clothes – the data pointed to here is very, very lacking. The anecdote was weak, but sufficient for what should have been an obvious task of poking holes in questionable data.
Instead, you prove my point by digging in your heels even more because someone tossed out an anecdote.