KeithS @475
Google has the advantage of a huge corpus of web pages with heaps and heaps of context and links. Amazon doesn't. How to figure relevance from non-standardized categories, blurbs, and otherwise patchy data is left as an exercise for the reader.
Funny, libraries pretty much figured this out ten years ago.
No, I'm not sympathetic to Amazon. They had a bone-headed agenda, they engaged in stupid IT practices, and have demonstrated similar cluelessness about customer relations and PR.
It is terribly interesting to compare Voyager, CIP, and British Library cataloging data for the same ISBN records at Amazon.
Jon #269
Well, no. Apparently, and it seems reasonable, everything in the database has an 'adult' flag, and the website simply reacts to whether an item's adult flag is Yes or No. (Which makes more sense to me than a 'display sales rank' flag. I would think a 'display sales rank' routine would instead check various fields in a catalog record to determine what to display.)
That means changing that tag in a lot of records could be as simple as doing a simple query to, say, update all records from id 100 to id 50000 setting adult=YES.
Except that isn't what happened. First, they did a records sweep on specific categories; their official statement pointing out that numerous subject categories "Gay & Lesbian themed titles--in fact, it impacted 57,310 books in a number of broad categories such as Health, Mind & Body, Reproductive & Sexual Medicine, and Erotica" make it clear that specific kinds of content were involved.
But the categories they used for their query are related to more specific content descriptors; see this discussion and the specific examples.
So they got their targeted records based on the categories they queried. Then they set the Adult flag that runs the script to update the visible sales rank data.
And it's a terminal interface, with lines showing the relationships between the tables. It's not a GUI. There's another tool, with a primitive GUI for multi-lingual stuff, but that would have had identical results on the French and German sites; the results are not identical except for the books in English with identical ISBN numbers to the US books. It's also interesting to look at the flagged categories in terms of why one edition was denuded of sales rank data when another wasn't.
You know, if Amazon actually apologized on their front page, and if they actually looked at the ethics of selling items that they don't want to acknowledge selling, I'd be less annoyed.
But now? I'm not buying the "French," or the "German" guy made a mistake--which, by the way, I've not seen from an Amazon PR person. And my friends who work there say that's not the way it works even now.
I am buying that the metadata for some books in inaccurate because the out sourced taggers don't speak English fluently, but that's not really the prime mover here.
The prime mover here is that Amazon wanted to hide adult content from search results, and, for whatever reason, they screwed up.
Larry@249
This is NOT a miss-click.
This is someone using GREP and SQL queries to first pull a set of records in the database and then to remove the flag that says "display sales rankings."
This was an employee doing what he was supposed to do, only doing it stupidly and badly.
I get that. I know, quite personally, how those things happen.
I also know that they could have posted a public apology by now. I've received two copies of the "ham fisted" email.
But I'm sorta thinking--I don't want to give business to a company that is this stupid about PR.
I'm also thinking I don't want a company that is this stupid about IT and rollbacks and re-tagging to have my credit card in its database.
I've just received essentially the same response from Amazon to my email query/complaint that the Seattle PI blog posted.
I still think they need an apology on the front page of their site.
And a Twitter account for Amazon PR.
It's not checkboxes. Really.
However; From the Seattle PI blog, quoting Amazon spokesperson Drew Herdener:
This is an embarrassing and ham-fisted cataloging error for a company that prides itself on offering complete selection.
It has been misreported that the issue was limited to Gay & Lesbian themed titles--in fact, it impacted 57,310 books in a number of broad categories such as Health, Mind & Body, Reproductive & Sexual Medicine, and Erotica. This problem impacted books not just in the United States but globally. It affected not just sales rank but also had the effect of removing the books from Amazon's main product search.
Many books have now been fixed and we're in the process of fixing the remainder as quickly as possible, and we intend to implement new measures to make this kind of accident less likely to occur in the future.
You can find the original here
This post on the specific category metadata pretty much satisfies me it was someone who in formulating the database query and commands to remove sales ranking, for whatever personal reasons, equated GLBT with "adult" or at least something to filter.
Dear Author on Amazon using category metadata.
One of the many things that puzzles me about this--let's assume it was in fact an effort to remove "adult" content from the top search hits and front page--
This is a stupid way to do that. The problem of defining adult has already resulted in prolonged trials. It's a matter of opinion.
The smart thing to do would be to allow users to decided what they do and don't want to see.
They already have user prefs; most of the infrastructure is in place.
Use third-party metadata from LOC and the British Library--data that Amazon already has in place and receives already--for the filtering settings, and let users decide what to filter.
There's just no sense in this snafu, at all. It's stupid all the way down.
Ginger @83: In this case, only non-abled, non-het sex were deranked, and the books actually affected were often not about sex (i.e., porn, etc.) but about human sexuality, non-hetero variety. A glitch doesn't contain embedded homophobia.
The largest common denominator of books with sales ranking removed is GLBT.
But het books--Alex Comfort's Joy of Sex--have been removed from sales rankings too, though not in all editions.
So while there's some heteronormative problems, quite a lot actually, with the definition of "adult," I really don't think there's an anti-queer corporate agenda.
I'm pretty sure about it. Jeff Bezos is a bookseller who wants to make money.
This kind of tactic is not going to make money.
I think it's bad data, bad decisions about glossaries and thesauri, and a poorly thought out process that started months ago, with a major roll-out on a nice quiet weekend.
When everything got snafu-d.
This edition of Lady Chatterley's Lover is not stripped of sales rank data:
http://www.amazon.com/Lady-Chatterleys-Lover-Cambridge-Lawrence/dp/0141441496/ref=ed_oe_p
There are lots of books where one edition is stripped of sales rank data, but another isn't. I've gone cross-eyed trying to figure out the differences, and honestly can't. Which is why I'm now leaning towards internal metadata, possibly customer supplied, in whole or in part.
I applied last year for a job at Amazon, but then decided not to take it. The flagging system result is ultimately passed to a human being who makes a yes this record should be permanently altered in this way, or not. If it needs to be altered the person writes a request describing the rationale and the change, and it then goes to someone with editorial access. I very much got the impression that Amazon liked live humans making decisions, and even more, that Amazon was in the business of selling books to make money.
This situation, while it is clearly based on poor decisions and poor metadata, was done en masse, via a sql query(ies). There are far too many books involved for it to have been done "by hand."
I don't think it's a "glitch" in the ordinary way. Honestly, at this point, having made myself dizzy looking at books and the associated metadata (not all of which customers can see, even in the HTML source) I think this is sort of what happened:
1. Corporate decides to bury "Adult" products by removing them from sales ranking.
2. Corporate and IT do not sufficiently identify the products' metadata--or define "adult." This is the first stage of WTF?
3. Actual live database SQL search is outsourced, and/or tagging procedure that preceded sql query and command to remove is done by outsourced non native English speakers. Second WTF? This might explain some of the very odd metadata for some books, and the odd spelling of some metadata.
4. I can't see why else books like "Heather Has Two Mommies" would be in the list of titles to be removed from sales rankings.
5. I do absolutely believe Corporate wanted to remove "adult" materials from front page lists. I don't believe glitch." I do believe stupid SQL queries and bad metadata. I don't believe Amazon has a list of "books to remove" compiled by humans, or that someone is clicking books individually. This was done directly to the database. I can't for the life of figure out what criteria for the metatdata--or even what metadata they used. My friends in Amazon editorial tell me they have a lot they don't show to us.
I don't think there's an anti-queer, anti-feminist, anti-disability agenda. Just human stupidity. Amazon wants to sell books, not change the world.
__________________
Allen @45 I'm a paid registered version person; but some of the features I depend on--templates and filters, are problems in the last commercial release.
And Linkmeister@49, exactly. I want the Eudora-built-by-Steve-using-standard-mbox-files and such.
I'm annoyed enough that I'm strongly considering going back to PINE and Vi or emacs, (no, no I don't care which).
That said, I still have to try Mailsmith from BareBones; I'm currently using mail.app; spent almost a year on Thuderbird, and it just hurt too much.
Hi
We seem to have a clean bill of health.
Go forth and post. We know we need to re-set the server clock, but since over 300 of you are now compulsively hitting re-load, we've turned on the forums.
Please do PM me, Medievalist, if you see something wonky.
Yes, of course, wonky is a technical term . . .
Thanks to all of you, and especially our ML hosts.
Y'all are finest kind.
OK.
We're at Stage II now. We've run the SQL utilities (thanks Scott), and we've now turned vBulletin off so we can run some other utilities, because the database got hosed (but it will be fine).
In the meantime, please please please stop clicking on Refresh and trying to log in. It slows things way down when, as just happened, 200 of you log in within five minutes of us getting the board up.
We appreciate your patience and enthusiasm, but please, just be patient a little longer. I'll be sure to come here, the IRC channel, and (not) Written in Stone to let you know when we're live.
Thanks guys; you're the best.
Lisa /Medievalist
We're running various utilities now; panic not. Just making sure the server has TLC.
Just hang in there peeps.
As Mac says, we have alternatives ready to do, including a second vBulletin license.
In the meantime, you can frolic at the AW sister forum: here
and Roger Carlson's AW Refugee Camp here:
here.
I've been dealing with multiple e-mail accounts since 1989. Accounts that routinely have upwards of 150 new mails, real mail, non-spam, non list mails, per day.
Don't delete the emails Patrick, but archive them somewhere you don't see them. Pick an arbitrary date, and archive everything before that. Send an automatic notice for the next two weeks or so that explains you've had to declare email bankruptcy.
People who are really important know other ways to reach you--they can trip you in the hallway, or call you on the phone.
Do it.
And iPhones are great for Grand Jury Duty. If you show them that you can turn off the phone part, you can use it to read e-books!
Be willing to do it again in a few years, if you have to.
OK.
I'll get in touch about seeing what we need to do to get y'all on MT 4.2 Pro.
Wow
This is so very very cool.
Congrats PNH and Tor folk; yes, I know, there are some glitches still, but this is really a pretty impressive undertaking.
I'm really excited.
I suspect the native language is Chinese, largely because of the oddities with tense and propositions. White Smoke reminds me of the way plagiarizing students will sometimes attempt to use Microsoft Word's thesaurus by right-clicking a word and selecting a random term from the thesaurus.
The irony is that the odd words they frequently choose serve to draw attention to their plagiarism.
| Year | Number of comments posted |
|---|---|
| 2009 | 17 |
| 2008 | 2 |
| 2004 | 1 |
Total: 20 comments. View all these comments on a single page.
The most recent 20 comments posted to Making Light by Lisa Spangenberg:
Show all comments by Lisa Spangenberg.