I find "anecdotes do not equal data" to be unscientific.
Your doctor doesn't take your oral history for fun, it's data. Now, just like a medical doctor we should weigh anecdotal data accordingly, but it's still data.
Here for example we can tell BMW/Audi reportedly sometimes make production errors. Tesla reportedly don't always make significant (to the commenter) production errors.
Sure, that's not terribly useful. It's still data.
The doctor takes your oral history to assess your health. Knowing that $x percent of family tree has high blood pressure means they can more accurately assess your risk of high blood pressure. For the population of “literally you”, your oral history is data. Now, if the next patient came in, and your doctor said “well, the last person in here has high blood pressure, you might have it to”, that’s an anecdote.
Similarly, telling somebody ~”My instance of $product worked great” isn’t really useful info for them. That’s a sample size of 1 out of a massive number.
Where only one or two anecdotes are presented, there is a larger chance that they may be unreliable due to cherry-picked or otherwise non-representative samples of typical cases
Your doctor doesn't take your oral history for fun, it's data. Now, just like a medical doctor we should weigh anecdotal data accordingly, but it's still data.
Here for example we can tell BMW/Audi reportedly sometimes make production errors. Tesla reportedly don't always make significant (to the commenter) production errors.
Sure, that's not terribly useful. It's still data.
/bugbear