Yes. Thank you for stating it so well. Self diagnosis is like a conjecture based on observation. It needs to be proved out formally to be a working theory. But that’s part of the process and there absolutely is value in self diagnosis. It’s up to professionals to sort the wear from the chaff, webmd be damned.
Similarly anecdotal evidence is just not strong evidence, but it is still evidence. Anecdotal evidence from a lot of people is a dataset and from a wide variety of people is a good dataset. It's all about confidence levels. Consider likelihoods (different from probabilities) and just update your priors but never forget the confidence intervals. They're the most important part!
