Doc Searls Weblog

10 responses to “What’s wrong with this assumption?”

1. 

   Tara Hunt

   November 5, 2009 at 4:06 pm

   You can become an author through author tools, but that doesn’t seem to affect the algorithm. I get recommended my book all of the time. :/

   Reply
2. Friends of Dave (friendsofdave) ‘s status on Thursday, 05-Nov-09 20:07:15 UTC – Identi.ca

   November 5, 2009 at 4:07 pm

   […] http://blogs.law.harvard.edu/doc/2009/11/05/whats-wrong-with-this-assumption/ a few seconds ago from api […]

   Reply
3. 

   William

   November 5, 2009 at 4:38 pm

   But Doc…I bet you DID look at it 🙂

   Reply
4. 

   Carter

   November 5, 2009 at 4:48 pm

   Isn’t that a teensy bit of an edge case? 😉

   Reply
5. 

   scritic

   November 5, 2009 at 4:54 pm

   Considering that there are so few authors and so many customers/readers, I’d say Amazon was justified in not adding a field where customers can indicate that they are authors of particular books.

   Reply
6. 

   Shawn Powers

   November 5, 2009 at 6:17 pm

   Irony aside, the algorithm seems to be working. There is a likelihood you’d be interested in the subject matter. 🙂

   Reply
7. 

   Morgan

   November 5, 2009 at 11:04 pm

   This seems like an edge case but it does point to a weakness in their overall recommendation engine, which is that it doesn’t learn enough about you even when there are clear opportunities to do so. For example I read a lot of poker books back in the mid 2000’s when it was all the rage. For a year period I probably bought 15 books. But for the last 3+ years I haven’t purchased or looked at a single one. Amazon continued to recommend them to me as recently as my last purchase. I can’t get away from the things. Now, with my recent binge on different types of titles the poker has finally left the radar; but it seems like Amazon has the opportunity to use the customer/author databases, the purchase frequency/time span and other more sophisticated elements of referral to our mutual benefit.

   Reply
8. 

   Brian Benz

   November 7, 2009 at 1:55 am

   You can work with them here:

   https://authorcentral.amazon.com/

   Reply
9. 

   gammydodger

   November 22, 2009 at 1:49 am

   Agree with Shawn, algorithm’s working well, but Amazon are unable to resolve user_id, author_name and person. Semantic definitions and technologies will resolve this soon enough allowing online applications to act with a little more intelligence.

   I just checked out Brian’s link to AuthorCentral, and looked at the source of http://www.amazon.com/Neil-Gaiman/e/B000AQ01G2/, but didn’t spot any RDF. Checked out wikipedia http://en.wikipedia.org/wiki/Neil_Gaiman, no luck there either.

   However, Freebase has the solution http://www.freebase.com/view/en/neil_gaiman – all machine readable. Doc, you’re there too: http://www.freebase.com/view/en/doc_searls

   If a little RDF was inserted here and there as we go, the data would all begin to join up nicely and these problems might go away.

   Reply
10. 

    Ron Reisman

    January 26, 2010 at 8:30 pm

    Sorry Doc, won’t happen unless they change their algorithm.

    Reply