[adjkant]

I think the part of the equation missing here is that you don't just have anecdata, you have data that more or less applies to most cases of your specific OS + Browser + other relevant customizations. If you have a highly customized system, you're simply more likely to experience issues from edge cases, yes. That data is also much less reflective of the average user.

No, it's not hard scientific data with a p-value. But IMO it's important to not simply write away all experience as anecdote not data without contextualizing the relevancy of the anecdata in question. I have seen way too many HN comments that look at usability and don't realize how specific their situation is that causes their issue. That and of course their complaints not being relevant/visible to the typical end user.