Real-time PR
Apr. 14th, 2009 11:25 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
jackThe history[1]
Over the weekend it was discovered that Amazon.com had de-listed a wide variety of books, prominently just about anything vaguely gay or lesbian, including children's books like "Heather has Two Mommies". This was a fairly major disaster: Amazon is a very important place to buy books, and removing books from turning up in searches practically makes them not exist. It began a month (??) ago. Authors said "I wrote this book, did a book signing, and several people said, I wanted to buy it on amazon, but couldn't find it." People found the extent to which the world tolerated their existence dropped a fraction, because they were gay.
There was a gigantic, gigantic shit-storm. There was much speculation. (I would say "uninformed speculation", but that's not true. Speculation was the only way real information was turned up.)
Customer service and PR had nothing useful to say, either giving an explanation for why actually obscene/adult books are removed at all, or (as of the weekend) that there was a "glitch", which without further background is not very convincing. Without better systems for percolating information through, that's inevitable, but it does raise the question of whether they SHOULD have better systems.
Now Amazon said it was their fuck-up and they were embarrassed. They are fixing it rapidly. But they didn't specifically apologise. Nor explain in any detail what had actually gone wrong. Some people apologised for thinking the worst.
[1] On the internet, "history" can refer to two days ago. (Or maybe a month ago.)
What went wrong
Ex-amazon employee and professional ranter Mike Daisey mentioned what he'd heard in slightly more detail than the press release, in the most recent couple of posts here
The facts seem to be:
1. The event was triggered by (according to Daisey's grapevine) human error by someone in France, which propagated everywhere.
2. There is a category "adult" which is automatically omitted from searches to prevent amazon being lambasted by people who come across something offensive accidentally. I'm not sure exactly how this works.
3. It seems said human error gave books in several meta-data categories were "adult" (reading between the very compact lines of the press release).
4. This could be a slip of finger executing a too-wide database update, a mistranslation of what things should be considered "adult", implementing a too-broad corporate policy, someone's personal prejudices intruding, etc, depending how amazon's database works, and I've no idea.
5. Probably (but not necessarily) someone mentally categorised books tagged as "gay" as adult, or something similar, which concatenated with further errors to lead to banning "heather has two mommies", etc.
6. It's not clear to what extent the human error is culpable, and to what extent having a system which can be broken in that way is culpable, and to what extent judging homosexual as adult is culpable. Was it a couple of prejudiced policies or meta-data which an error exposed, or mostly a combination of minor technical and communication mistakes[2].
The extent to blame amazon depends on what happens and what you think is unnacceptable. Blame them for unlisting books ever? Maybe. Blame them for deliberately suppressing books on certain topics? Maybe. Blame them for having imperfect database policies that can cause this sort of snafu, maybe no, maybe yes, maybe yes but not that much. Etc.
I feel in a very uncomfortable superposition, in that I need to either be very very outraged, a little bit outraged, or not outraged, and I don't know which. And eg. if I'm a little bit outraged, it might either look like (a) I think amazon were only a little bit culpable or (b) I only care about discrimination a little bit.
[2] Notice, it might be mostly A's fault, slightly B's fault, or slightly A's fault, slightly B's fault. The latter case has less fault but caused just as much problem: that's what we call a "fuckup", but it's arguably better than malice.
Lessons
Hooboy. Are there some lessons. Even though I still don't know what actually contributed to the error, there are lots of things to observe, that will go on being true even if this problem is resolved.
1. PR is getting faster and faster. We've had this before, with telegraphs and things, when all of a sudden people had to be responsible for their actions in weeks not years. But Amazon is reasonably practised at this, and still not able to handle such a crisis. Many companies are totally unable to deal with a major PR crisis on a weekend, when everyone has already boycotted them by Monday. Even a coherent "help! I don't know. someone will issue a full statement on tue" would help a lot.
2. Amazon, Livejournal, Google and others have become fundamentally important to varyingly vast numbers of people. Not always life-or-death important, but a small whim from the company can impact a large number of people. Before, we have relied on nationalisation, oversight committees, free-market competition and other strategies to make companies behave. For good reason: if your gas company fucks you over, you have to wait for EVERYONE to get fucked over before it's economical to found another one. Much more efficient to band together and insist they get it fucking right the first time. We are in a similar situation here, except that with the web such a quick world, we have the chance to demand transparency, yet are not sure what measures are sensible.
3. Databases and meta-tagging systems etc for gigantic websites are complicated things. This sort of thing is going to go on happening.
4. This was a big twitter event. Whatever twitter is for, this is presumably part of it.
Links (some out of date)
Amazon calls mistake 'embarrassing and ham-fisted'
http://dearauthor.com/wordpress/2009/04/12/amazon-censors-its-rankings-search-results-to-protect-us-against-glbt-books/ (originalish whistle-blow
http://markprobst.livejournal.com/15293.html
http://tehdely.livejournal.com/88823.html (Analysis, and was it hacked.)
http://nielsenhayden.com/makinglight/archives/011173.html#011173 (Making light on the topic. I shudder to see Kramer there, and continuing fallout from racefail, although apparently well moderated.)
Over the weekend it was discovered that Amazon.com had de-listed a wide variety of books, prominently just about anything vaguely gay or lesbian, including children's books like "Heather has Two Mommies". This was a fairly major disaster: Amazon is a very important place to buy books, and removing books from turning up in searches practically makes them not exist. It began a month (??) ago. Authors said "I wrote this book, did a book signing, and several people said, I wanted to buy it on amazon, but couldn't find it." People found the extent to which the world tolerated their existence dropped a fraction, because they were gay.
There was a gigantic, gigantic shit-storm. There was much speculation. (I would say "uninformed speculation", but that's not true. Speculation was the only way real information was turned up.)
Customer service and PR had nothing useful to say, either giving an explanation for why actually obscene/adult books are removed at all, or (as of the weekend) that there was a "glitch", which without further background is not very convincing. Without better systems for percolating information through, that's inevitable, but it does raise the question of whether they SHOULD have better systems.
Now Amazon said it was their fuck-up and they were embarrassed. They are fixing it rapidly. But they didn't specifically apologise. Nor explain in any detail what had actually gone wrong. Some people apologised for thinking the worst.
[1] On the internet, "history" can refer to two days ago. (Or maybe a month ago.)
What went wrong
Ex-amazon employee and professional ranter Mike Daisey mentioned what he'd heard in slightly more detail than the press release, in the most recent couple of posts here
The facts seem to be:
1. The event was triggered by (according to Daisey's grapevine) human error by someone in France, which propagated everywhere.
2. There is a category "adult" which is automatically omitted from searches to prevent amazon being lambasted by people who come across something offensive accidentally. I'm not sure exactly how this works.
3. It seems said human error gave books in several meta-data categories were "adult" (reading between the very compact lines of the press release).
4. This could be a slip of finger executing a too-wide database update, a mistranslation of what things should be considered "adult", implementing a too-broad corporate policy, someone's personal prejudices intruding, etc, depending how amazon's database works, and I've no idea.
5. Probably (but not necessarily) someone mentally categorised books tagged as "gay" as adult, or something similar, which concatenated with further errors to lead to banning "heather has two mommies", etc.
6. It's not clear to what extent the human error is culpable, and to what extent having a system which can be broken in that way is culpable, and to what extent judging homosexual as adult is culpable. Was it a couple of prejudiced policies or meta-data which an error exposed, or mostly a combination of minor technical and communication mistakes[2].
The extent to blame amazon depends on what happens and what you think is unnacceptable. Blame them for unlisting books ever? Maybe. Blame them for deliberately suppressing books on certain topics? Maybe. Blame them for having imperfect database policies that can cause this sort of snafu, maybe no, maybe yes, maybe yes but not that much. Etc.
I feel in a very uncomfortable superposition, in that I need to either be very very outraged, a little bit outraged, or not outraged, and I don't know which. And eg. if I'm a little bit outraged, it might either look like (a) I think amazon were only a little bit culpable or (b) I only care about discrimination a little bit.
[2] Notice, it might be mostly A's fault, slightly B's fault, or slightly A's fault, slightly B's fault. The latter case has less fault but caused just as much problem: that's what we call a "fuckup", but it's arguably better than malice.
Lessons
Hooboy. Are there some lessons. Even though I still don't know what actually contributed to the error, there are lots of things to observe, that will go on being true even if this problem is resolved.
1. PR is getting faster and faster. We've had this before, with telegraphs and things, when all of a sudden people had to be responsible for their actions in weeks not years. But Amazon is reasonably practised at this, and still not able to handle such a crisis. Many companies are totally unable to deal with a major PR crisis on a weekend, when everyone has already boycotted them by Monday. Even a coherent "help! I don't know. someone will issue a full statement on tue" would help a lot.
2. Amazon, Livejournal, Google and others have become fundamentally important to varyingly vast numbers of people. Not always life-or-death important, but a small whim from the company can impact a large number of people. Before, we have relied on nationalisation, oversight committees, free-market competition and other strategies to make companies behave. For good reason: if your gas company fucks you over, you have to wait for EVERYONE to get fucked over before it's economical to found another one. Much more efficient to band together and insist they get it fucking right the first time. We are in a similar situation here, except that with the web such a quick world, we have the chance to demand transparency, yet are not sure what measures are sensible.
3. Databases and meta-tagging systems etc for gigantic websites are complicated things. This sort of thing is going to go on happening.
4. This was a big twitter event. Whatever twitter is for, this is presumably part of it.
Links (some out of date)
Amazon calls mistake 'embarrassing and ham-fisted'
http://dearauthor.com/wordpress/2009/04/12/amazon-censors-its-rankings-search-results-to-protect-us-against-glbt-books/ (originalish whistle-blow
http://markprobst.livejournal.com/15293.html
http://tehdely.livejournal.com/88823.html (Analysis, and was it hacked.)
http://nielsenhayden.com/makinglight/archives/011173.html#011173 (Making light on the topic. I shudder to see Kramer there, and continuing fallout from racefail, although apparently well moderated.)
no subject
Date: 2009-04-14 02:27 pm (UTC)The consistency of the book removal with the tags seems to make it true that books were removed by their metadata, but doesn't explain why some books were removed earlier and then more later. That looks a lot more like a policy being implemented and taking time to be more largely implemented than it looks like a mistake.
The policy is too stupid to make sense as an actual policy, so I buy that some mistake happened between the start of the policy and this implementation, but it sure does look like initially some books with homosexual content were removed deliberately when similar books without homosexual content were not.
Which is part of why I await an explanation from Amazon. I don't feel any explanation will be satisfying until it also explains why they told people different things before they told us whatever they're currently telling us.
no subject
Date: 2009-04-14 04:49 pm (UTC)