> It might be true, but this attitude makes it very hard for developers not to discount all complaints, including ones grounded in reality.
I think this is a problem for outside observers, not competent developers. Unlike us, Apple developers have ready access to statistics about bug reports and other forms of support requests that we do not. This data, presumably, is not subject to the same biases that infect an anecdote-driven discussion of software quality by outsiders.
In my experience, you get a bug report for every hundred or thousand affected users, if that; so excuse me for not having much faith in such statistics.
> This data, presumably, is not subject to the same biases
... but it's likely subject to many other biases, e.g.
* "hey, this bug was reported by iLife devs, better prioritise ahead of that bug that has affected millions of users for almost two years -- them people are not going to shout at me in the canteen."
* "hey, this bug was very well-reported by very technical server people, let's prioritise it ahead of that bug affecting millions of semi-literate consumers"
* "hey, this bug blocks the release of $shiny-new-iPhone-feature, let's prioritise it ahead of that bug affecting the trackpad of penny-pinching laptop users"
I wasn't talking about the biases that affect a team's response to a given bug but, rather, the ability merely to know whether a given release has more bugs or fewer bugs than others, and the biases that can effect this count.
> In my experience, you get a bug report for every hundred or thousand affected users, if that; so excuse me for not having much faith in such statistics.
But the number of reports should still be roughly proportional to the number of bugs in the wild, shouldn't they? That's all that is necessary to compare one release to another, particularly if all you're looking for is significant quality degradation, on the scale discussed in the article. And is it really your position that these statistics are worse than a few users' anecdotal views about which release is better?
> But the number of reports should still be roughly proportional to the number of bugs in the wild, shouldn't they?
That's just an article of faith. Regardless, part of the problem is where those bugs are. Maybe file-tagging involves millions of LoCs and it's now completely bugfree, but if wifi connections keep having DNS problems because of one single bug, overall experience is much more affected than it would be in the opposite case (bugfree wifi and buggy file-tagging).
> And is it really your position that these statistics are worse than a few users' anecdotal views
When "a few users" are your most ardent evangelists (Arment, Gruber etc), I'd say you should worry regardless.
I think this is a problem for outside observers, not competent developers. Unlike us, Apple developers have ready access to statistics about bug reports and other forms of support requests that we do not. This data, presumably, is not subject to the same biases that infect an anecdote-driven discussion of software quality by outsiders.