April 13, 2009

The End of Offense?

I suspect that someone, somewhere is sitting back, being very smug about the internet eruption that is amazonfail. They think they've shown teh gayz and teh wiminz. They haven't got a clue what they've just done.

If you don't know what I'm talking about, start at Sex in the Public Square, which is just generally a good place to start. Caroline's got a good rundown of the situation in which books with non-explicit GLBTQ romances, nonfiction books on "alternative" sexuality (including sexuality for the disabled), and feminist theory were listed by Amazon as adult and stripped of their sales ranks. This means they didn't show up in searches, making them very difficult to buy.

Aside from issuing a statement that the removal from sales rankings was a glitch and is being fixed, Amazon has been very quiet about the whole problem. Several people have pointed out that the selective list of titles involves makes any purely technical explanation vaguely ridiculous. Over on LiveJournal, however, tehdely offers a theory:

It's obvious Amazon has some sort of automatic mechanism that marks a book as "adult" after too many people have complained about it. It's also obvious that there aren't too many people using this feature, as indicated by the easy availability (and search ranking) of pornography and sex toys and other seemingly "objectionable" materials, otherwise almost all of those items would have been flagged by this point. So somebody is going around and very deliberately flagging only LGBT(QQI)/feminist/survivor content on Amazon until it is unranked and becomes much more difficult to find. To the outside world, this looks like deliberate censorship on the part of Amazon, since Amazon operates the web application in question.

Okay, there's no certain information there, but the idea does make it a little easier to comprehend a WTF situation. Combine concerted action by an outside party with a naive (but fairly standard) and probably automated offensive content policy from the company in question, and you've got instant censorship with no action on the part of the company. In fact, Amazon would even almost be right to consider this a glitch.

The problem for the crusading censors, though, is that this isn't a glitch. This is a system set up to run on trust, and it's been gamed. And in abusing that trust and gaming that system, the censors have cost Amazon badly. Net Effect summarizes the costs to Amazon in terms of publicity so far and in the likely near future, but there are other very important considerations for Amazon.

The company has been working with authors to make Amazon the place to promote their books. Associate accounts give authors a small amount of money on every sale made with one of their links, encouraging them to push readers to Amazon. Amazon makes preorders (which can mean a lot when an author is negotiating the next contract) very simple and available earlier than nearly any other seller. And Amazon offers authors' blogs, fora for writers to interact directly with people who are considering buying their books.

All of those things are great for authors and for Amazon. None of them will make a bit of difference if an author doesn't trust Amazon to keep random whackos from arbitrarily making their books disappear. Authors will find somewhere else to promote themselves.

In other words, the censors are messing with Amazon's money, which changes the game. YouTube has had very little to lose by removing videos tagged as offensive. Flickr has had very little to lose by deleting entire portfolios when someone complains that one picture should have been labeled adult but wasn't. Same with WordPress delisting adult content. With a free service, someone gaming offensive content policies generally only costs the person whose material is removed.

This is different. Amazon has a huge incentive not to be gamed. They need to fix this and quickly. Considering the sales figures of some of the books affected, like Brokeback Mountain and Ellen DeGeneres's autobiography, they have an incentive to fix it by telling the censors to piss off. They can't sell books no one can find.

I wrote last year about the idea of offensiveness and questioned whether it was still a useful concept in a diverse, egalitarian society. This is an excellent example of what I was talking about. What does Amazon do when one group is offended by content and another is offended by censorship? They do what Amazon does--sell books.

I'm hopeful that amazonfail will be the beginning of the end for offensive content policies. I definitely think it will lead toward the elimination of any automated systems for dealing with complaints of this kind at commercial ventures. Considering the cost of dealing with complaints on a case-by-case basis, it's quite reasonable to conclude that this will, in general, result in a liberalization of content. It's just so much easier to say, "We don't censor," than, "We'll look at that book and make a decision. Yes, and that book too. And that one. And...."

It particularly easier when Amazon already has a system in place for showing people what they think they want them to see. "Yes, sir. Just log in and you can click on a button next to any content that offends you. As long as you're logged in, you won't see it anymore."

There is every reason to think that the cost of incidents like amazonfail will push technology companies to make users more responsible for the content that they see while leaving the rest of us alone, which is exactly the opposite of what the censors who are so cluelessly smug today were hoping for. Poor little moralists, but they should have seen it coming.

After all, it's the rest of us who read.

Update: See today's post for Amazon's statement on what actually happened.

11 comments:

Anonymous said...

Of course, none of them read. If they did they would know what was in that bible of theirs.

Stephanie Zvan said...

I was very close to going there in the post. :) In the end, I decided that it was all equally true for any other type of moralist, too, so I didn't.

Peggy K said...

While it's looking like the deranking of books was likely due to incompetence rather than malice (see here), I'm hoping that the whole FU does result in a user-driven preference system like you describe. Two things that became clear to me is that "offensive" is in the eye of the beholder, and there is no easily automated way of determining whether such content is present in a book unless you eliminate broad categories (e.g., "anything remotely related to positive depictions of homosexuality").

I'd like to think that the massive response to Amazon's fail will have a positive impact. It's like the mirror image of one of those Focus on the Family anti-gay letter-writing campaigns.

Stephanie Zvan said...

I'm hoping so, too, Peggy. I'm also hoping that more people settle down a bit (not stopping demands that this be fixed, but cutting out the "Amazon has told us three things that are not exactly identical; there must be a conspiracy!") before Amazon decides that they've already lost with this group and have nothing to gain by being good citizens. I think it's happening, but the continuing froth disturbs me.

I also can't see any way to stop this from happening in the future without a user-driven system. Anything that is more universal will be subject to the same problems they're encountering now. I'm sorry for the people who were delisted and felt threatened, but I'm almost glad it happened when it did. This is a weakness in the model that needed to be exposed.

DuWayne Brayton said...

There exists a definite temptation to be an ass and just whack a lot of religious material into the same oblivion, but I do hope that doesn't actually happen. I can definitely see very reasonable justifications for it - some that would indicate utility beyond catharsis. But the bottom line is that it would make "us" no better than "them."

This (well, actually Greg's post) is the first I heard of all of this. I am definitely going to write Amazon about it and make it clear that I want to continue to use them (which I assuredly do) as a customer and at some point as an author, but that this policy would definitely prohibit me from doing so.

As an aside, I think a very good alternative to using this sort of generalized system, would be to allow organizations to create amazon fronts that allow them to do the same sort of flagging, but only for people who enter through that organization's portal. At that point people who find sanctimonious prickery offensive, could make that harder to find and people who are terrified that their children (or husband/wife) might learn about Teh Gay, can be sanctimonious pricks.

Ahhh, so many things to blog about and not nearly the time. Though semester ends soon and I have a couple weeks - I just hope I don't fill all that time with other stuff.

Stephanie Zvan said...

DuWayne, this situation is going to require at least as much attention after we fully understand what happened as it does now. I wouldn't worry about having to wait. You'll just end up with a more informed post. This one is outdated already (see Peggy's link).

I like the idea of a portal. If Amazon runs it, it might even not die like the last one. However, if the content is user-controlled for a large group of people, you will see anti-censorship folks getting in there and gaming the system. Most of them will be nineteen, and it will be a bad thing to do, but it will happen.

DuWayne Brayton said...

However, if the content is user-controlled for a large group of people, you will see anti-censorship folks getting in there and gaming the system.That could be easily dealt with by putting actual controls in the hands of the organizations, rather than the general public.

AK said...

Two comments:

First, IMO most people who find stuff "offensive" are more interested in keeping other people from seeing it than themselves. They use various excuses (especially regarding children), but they're really closet Nazi's.

Second, I have many years' experience with software development, especially when it comes to user-maintained tables and other global parameter systems. The only solution I know of that's generally reliable is to set up a test system for user maintenance, a separate test system from the one used for software upgrades. This is very expensive, both in terms of software/hardware, and user time/training. Worse yet, it never gets included in the up-front cost estimates.

Amazon can certainly afford to set such a thing up, I'm disappointed (but not completely surprised) they didn't do it up front, or at least in response to some less glaring error (there must have been a few). But I doubt most of the sites that aren't selling anything (besides advertising) will bother to pay for it, even if it's available in whatever packages they're using.

Dr. A said...

thanks for this great post and the links. fail indeed.

Stephanie Zvan said...

AK, the more I know about how Amazon runs their systems, the more it sounds as though they're held together with baling wire and twine. It's actually a little scary.

AK said...

I've never actually looked into the systems at Amazon, but as a customer my impression has been that they were very robust and well-designed compared to most of the software out there.

The biggest problem, IMO, is that these new, very successful, companies grow too fast too keep their corporate culture intact. They bring in a lot of professionals experienced in the traditional way of doing things, and end up with all the problems of the more established companies.

Microsoft (IMO) is an early case in point, and Amazon is probably another. I would expect that in the last decade, they've gone from an innovator in systems development philosophy to a carbon copy of most other big companies (and government agencies), with most of the same problems.

The biggest problem with Internet software is quality control. There are two reasons for this (at least):

1. Most Internet software development came out of the PC environment, where quality control standards were very late to develop. Good quality control adds tremendous costs to a project, which usually aren't factored in up-front.

2. Internet quality control standards are usually modeled off of the older mainframe-based quality control, but the environments are very different and risks that are negligible with mainframe applications can be critical with Internet applications. Every application has to be retested with each new browser release, on a schedule that the application owner can't control.

Of course, these don't seem to be the problem here, instead it's a matter of understanding the risks associated with driver-type tables and their maintenance. I suspect what happened was that the issues with a user making a mistake while maintaining a table was brought up, and the systems people said "it's not our problem: make sure the users don't make mistakes". This is the traditional systems response to such issues, and while Amazon may have been innovative enough to deal with it early on, by now I suspect their systems departments have been so infected with the tradition that they did just what every other company I've ever heard of did (in this regard): ignored the problem.

See also my post at Greg Laden's Blog.