I really wanted to comment on <i>Brandon Sanderson, fame, and the failure of the Long Tail</i> but alas comments on that blog are restricted to paid subscribers (=? uncritical fans) so any feedback the author gets will be just a tad biased.
What he's missing is that Amazon used to produce excellent book recommendations. They eventually "fixed" that - last I checked they don't even have an accurate search function, preferring sponsored results to accurate ones. The bad recommendations are a conscious choice, to extract the maximum possible money over the short term, rather than satisfying their potential customers.
LibraryThing still produces fairly decent recommendations, based on the aggregate of what a particular user owns. They may be a bit too heavy on bestsellers, but nothing like what Claxton describes.
Yeah, I've noticed the decay of the recommendations on Amazon as well, but I suspect that might be a function of the scaling argument that he's making in the piece. That is, when Amazon was relatively small and used largely by people trying to find books they couldn't get in their local stores, it was more likely for people who bought any particular book to also have bought some quirky stuff, and that would diversify the recommendations somewhat. As it's become bigger, it's the primary book source for more and more people who mostly just buy bestsellers, so those overwhelm everything else and are the only things that show up in the recs. Something like LibraryThing remains more niche, and thus can do a better job of avoiding the really obvious stuff.
I recall a big change, that seemed sudden to me - not in the recommendations, but in the search. It used to be, if I searched for an author's name, I'd get a list of the books they had by that author, and perhaps also books about that author. But one day when I searched for an author, I got books in the same genre, mostly by other authors. Some but not all of the incorrect search results were marked as "sponsored".
My conclusion: Amazon expects to get more money by showing me books I'm NOT looking for, than they expect to by showing me what I actually requested. Presumably they are paid a lot per unwilling eyeball, enough to offset any lost sales. Perhaps they figure the average customer will wade through the advertisements, labelled and unlabelled, and buy what they intended in spite of the spammage. Or perhaps they figure the average customer is likely to actually buy one of the advertised books then or later, and unlikely to react to the poor search by NOT buying the book they actually wanted.
I'm not their target customer, apparently. My time is too valuable to me to waste it trying to find products to purchase in spite of an intentionally broken search. Barnes & Noble had, at the time, a functioning search feature, and prices that might be higher than Amazon, but not by much... on average, not by as much as I valued lack of aggravation and wasted time. So I formed the habit of going to B&N first.
But the point I'm making here is that this behaviour has nothing to do with scale. It is related to market dominance - one of 100 alternatives simply cannot get away with abusing their customers, larger of 2 or 3 alternatives apparently can.
As for recommendations, they simply dropped their original feature, and replaced it with what they have today. Perhaps Goodreads was supposed to replace the original feature, once it became fully owned by Amazon.
I really wanted to comment on <i>Brandon Sanderson, fame, and the failure of the Long Tail</i> but alas comments on that blog are restricted to paid subscribers (=? uncritical fans) so any feedback the author gets will be just a tad biased.
What he's missing is that Amazon used to produce excellent book recommendations. They eventually "fixed" that - last I checked they don't even have an accurate search function, preferring sponsored results to accurate ones. The bad recommendations are a conscious choice, to extract the maximum possible money over the short term, rather than satisfying their potential customers.
LibraryThing still produces fairly decent recommendations, based on the aggregate of what a particular user owns. They may be a bit too heavy on bestsellers, but nothing like what Claxton describes.
Yeah, I've noticed the decay of the recommendations on Amazon as well, but I suspect that might be a function of the scaling argument that he's making in the piece. That is, when Amazon was relatively small and used largely by people trying to find books they couldn't get in their local stores, it was more likely for people who bought any particular book to also have bought some quirky stuff, and that would diversify the recommendations somewhat. As it's become bigger, it's the primary book source for more and more people who mostly just buy bestsellers, so those overwhelm everything else and are the only things that show up in the recs. Something like LibraryThing remains more niche, and thus can do a better job of avoiding the really obvious stuff.
I recall a big change, that seemed sudden to me - not in the recommendations, but in the search. It used to be, if I searched for an author's name, I'd get a list of the books they had by that author, and perhaps also books about that author. But one day when I searched for an author, I got books in the same genre, mostly by other authors. Some but not all of the incorrect search results were marked as "sponsored".
My conclusion: Amazon expects to get more money by showing me books I'm NOT looking for, than they expect to by showing me what I actually requested. Presumably they are paid a lot per unwilling eyeball, enough to offset any lost sales. Perhaps they figure the average customer will wade through the advertisements, labelled and unlabelled, and buy what they intended in spite of the spammage. Or perhaps they figure the average customer is likely to actually buy one of the advertised books then or later, and unlikely to react to the poor search by NOT buying the book they actually wanted.
I'm not their target customer, apparently. My time is too valuable to me to waste it trying to find products to purchase in spite of an intentionally broken search. Barnes & Noble had, at the time, a functioning search feature, and prices that might be higher than Amazon, but not by much... on average, not by as much as I valued lack of aggravation and wasted time. So I formed the habit of going to B&N first.
But the point I'm making here is that this behaviour has nothing to do with scale. It is related to market dominance - one of 100 alternatives simply cannot get away with abusing their customers, larger of 2 or 3 alternatives apparently can.
As for recommendations, they simply dropped their original feature, and replaced it with what they have today. Perhaps Goodreads was supposed to replace the original feature, once it became fully owned by Amazon.