More on #amazonfail

Okay, so as of today the prevailing winds of opinion on the Net appear to be hitting three major points:

  1. It’s nice that Amazon did acknowledge their error, although from scattered reports I’m picking up, not everybody has had their previous sales rankings restored. I have not yet been able to confirm any specific de-ranked books that haven’t been restored, myself.
  2. It’s not nice that Amazon hasn’t actually come right out and said “we’re really, really sorry about this, yes, this was a screwup of the highest order”. I’ve seen at least one author outright demanding an apology since her writing is her only source of income, and Amazon screwing this up therefore adversely affected her.
  3. It’s also not nice that the mechanism for hiding items globally is there to begin with. Charles Stross ably expressed concerns about this over here.

I’m of two minds about it in general, myself. One, having worked for places with large web-facing databases that can blow up in spectacular fashions if you look at them funny, I have every sympathy for the thought that some hapless employee somewhere might simply have blown something up. Also, based on my own workplace experience, it makes sense to me that even if Amazon’s quickly re-ranked all the de-ranked books, it’ll take a bit for all the fixes to re-populate out into the system. They are dealing with massive numbers of parallel servers, and massive amounts of data that all needs to get sent out to the server clusters. This takes time.

On the other hand, like Mr. Stross, I’m bothered by the fact that the capability to unilaterally hide anything tagged “adult” is there to begin with, as well as the fact that this got unilaterally applied to so many thousands of books that should never had been thusly tagged.

Long and short of it for me is, I’m not looking for the heads of Amazon execs on pikes at this point, but I do remain in a state of concern. The initial cranky mail I’d thought of sending, and which I held back until such time as we got a statement out of them, shall I think turn into an expression of deep concern. I’ll have to think about what to say.

Tonight’s further reading on the topic:

And of especial note, pulled out of the comments on the Scalzi post, is this one:

4 Replies to “More on #amazonfail”

  1. I gotta agree. Looks like Amazon got played like a lute, in a fashion difficult to prove especially if you want to keep proprietary code out of the public record… I am concerned that the speaking parts of Amazon didn’t treat this the same way Engineering did, e.g. Sev1, worth waking people up on Easter Sunday…. and that somebody obviously thought that unmoderated deranking was a Good Idea (or at least, didn’t think it through)….

    The word that comes to mind is “sloppy”. I don’t care for sloppy.

    1. Technoshaman @ 1: “Amazon got played” implies that someone did this deliberately, a supposition for which we have no real evidence; in fact, the one public claim of having done this deliberately has already been debunked.

      It also goes against their own statement of “embarrassing, ham-fisted cataloguing error”, for what it’s worth. We have no solid evidence or even a solid statement out of Amazon on what that actually means. But as I said in another comment to Paul on this thread, since I’ve worked for places with large customer-facing databases, I can definitely conceive of how easy it is for one person typing in a query to fuck it up catastrophically. In the absence of any actual data as to the motive of the person who caused the cataloguing error in question, I’m willing enough to chalk it up to “horribly, horribly unfortunate screwup” and move on. And hope that Amazon will take steps to prevent any such massive cataloguing errors in the future.

    1. Paul @ 2: I’d heard about this, yeah, and fair ’nuff. I’d be more comfortable with the system if it were exclusively user-driven, though. Some sort of check does need to be in place to make sure that huge swaths of products don’t suddenly vanish out from the sight of people who might actually want to, y’know, buy them.

      But as I said–having worked for places with large web-facing databases, yeah, I can conceive of how this could have been a particularly catastrophic error. Thanks for chiming in!

Comments are closed.