I'm not claiming that no one at Amazon was aware that the issue had public visibility. I'm not even claiming that the public visibility in this case didn't contribute to the swift, clear, and visible resolution we've seen.
Instead, I'm asking for evidence that high visibility correlates strongly with a resolution. Honestly, I will not be surprised if it exists. But no number of highly visible stories that are resolved will provide that evidence on their own.
Certainly, the way in which a large company discovers that an issue has become highly visible is completely irrelevant. The question is whether, if an issue doesn't become highly visible, do they still try to resolve it or do they ignore it?
Based on the email she got I'd say from Amazon's point of view the matter was closed. There had to be some prod to get them to take another look. If not the publicity, what was it?
Somehow, this is less clear to others than it is to me. I'll break it out into some cases. If the data points are mostly
- a mix of "publicity -> resolution" and "no publicity -> resolution", then publicity is irrelevant.
- a mix of "publicity -> resolution" and "no publicity -> no resolution", then truly the moral of the story is to have publicity if you want a resolution.
- a bunch of "publicity -> resolution", then we cannot make any conclusions!
We hear about the issues with publicity because, well... they have publicity. We make up the other half because it fits our preconceived bias.
Which by definition we don't hear about. In fact, that's why - if Amazon are rational - they'd give special treatment to issues that get publicity, so it seems like they handle customer issues better. There is essentially no downside for them in doing this; it's not like they'll be hurt by the publicity backlash from incidents that went unresolved because they got no publicity.
This story bubbled up through HN, BoingBoing, and many other tech news outlets. Someone at Amazon inevitably saw it. Nobody's claiming they randomly stumbled across the original blog entry.