![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
danieldwilliam) wrote2012-05-15 11:36 am
On Improving The Curation of Amazon's Book Offering
I’ve recently bought a Kindle and so has my Mum. In an attempt to reduce both my waistline and my monthly expenditure on lunch I’ve started eating soup in the canteen and using the saved money to buy ebooks. On a number of occasions I’ve logged on to Amazon and been overwhelmed by the choice. Overwhelmed to the point of paralysis. I’ve gone to the shop with an open wallet then walked out without spending any money.
I want a system that combines knowledge about my tastes with better meta-data about books on offer and a reputation management system for reviews and reviewers.
Here are some suggestions for Amazon about how they could help me spend money.
Nine Star Reviews
I personally don’t find 5 star reviews particularly useful. Is a 5 star book the best book ever written or just a really good book? 9 stars I think gives a more graded response.
Review the Reviewer and the Review
There are a number of reviewer on Amazon who write well considered, thoughtful reviews that help me understand if I would enjoy a book and if it is well written. There are also reviewers who don’t write reviews like that. I’d like the ability to review the reviews. I know Amazon already provide a “Was This Review Helpful” button. That’s a good start but I’d like to know if the review was really very helpful, or just a little bit more useful than a poke in the eye. I’d like to see the assessments of review flow through into an assessment of the reviewer. That way I could look at a review they had written that hadn’t been marked but see that generally this reviewer tends to give helpful reviews and therefore decide to read their review of a particular book.
I’d like to be able to ignore certain reviewers. Either because they are poor at reviewing, we have very different tastes or fundamentally different world views. (I recall looking at a book about a particular type of child discipline favoured by certain types of Christian conservative and described as child abuse by other people. Anyone who gave that book a high star rating I don’t want to read anything they liked.)
Tagging.
I’d like to see much greater use of tagging as a way of crowd sourcing meta-data about books. I’d like to be able to “like” tags and “not like” tags. (Please Amazon, no more vampires for me.) When people tag books I’d like to able to rate that tagging. Is Lords and Ladies a book about Shakespeare – you be the judge. Charles Stross has something to say about the death of genre and it’s replacement with tag-clouds. I’m not convinced but that’s really for Amazon to decide when they decide how much effort they, or their readers, are going to put into curating their offering.
Links to “Pro” Reviews.
I’d like to see Amazon reach outside the Amazon community and link to reviews of books. Not just professional reviews but any internet review. Again, I’d like the opportunity to rate these reviews.
The very fact that someone has bothered to write a non-Amazon review persuades me that this is a proper book and not a work of self-published vanity that should be at the bottom of a slush pile somewhere.
My Library
I’d like a quick and easy way of uploading my library to Amazon so I don’t have to keep correcting the recommendations. What I’d really like would be a bar code scanner on my iPhone or better, the ability to take a photograph of my bookshelves and for an App to work out what I’ve got from the spines. (This would require some way of tagging books as Mine or MLW’s or Bluebird’s or The Captain’s) I don’t mind if this allows Amazon to know a lot about me and then try and sell me books it thinks I would like to buy. That’s what I want. It’s easy to sell me the next Lindsay Davies book once you’ve spotted that I own all the others but it’s not helpful to me and I already know I like Falco. What’s helpful to me is being to see what I like from what I own and then find books and authors I didn’t know I liked until you suggested them. That’s what will prevent me logging on to see if the latest Falco book is in paperback yet and then switching off without buying anything when it isn’t.
Weighting of Reviews
We all know that 10’s from Alesha don’t count. I’d like to see Amazon do three things.
Firstly, normalise the reviews people give. If one reviewer has only ever given 1 rating of a 9 I’d like that to reflected when I see they have reviewed a book as a 7. This could be going on behind the scenes. It could be on the surface (perhaps a side note that says “1% of this reviewers reviews are a 1 – they must have really thought this book sucked to give it a 2”.
Secondly, weight the reviews of people who I think have good taste higher for me.
Thirdly, for written reviews I’d like to see some kind of analysis of the words that will help determine if a reviewer has put any thought in to the review. “Really good” would score worse than “Really good because…”
Matching Me With Reviewers with Similar Tastes.
I’d like to be able to see people who have bought similar items to me and then given them similar scores and then see what they’ve bought that I haven’t. Clearly there are some issues with privacy here. I don’t need to see that Jane Smith of Wistful Lane, Pergamon liked several books I liked and has just bought 50 Shades of Grey and had it shipped to her work address. I would be interested to see that ten people who bought and liked Cryptonomicon and Accelerando and all the Louis de Berniers magical realist books also bought and liked Singularity Doorstop in the Andes – Part Deux.
Play Lists
I’d like to see user created reading lists, akin to Play Lists. I’d like to a list of Books about Robots, Books about Post-Mortality, Books with Rubbish Endings, Books with a Strong Partnership, Books with a Lesbian Protagonist, Books to Read Once You’ve Read JK Galbraith, Books to Read If You Want a Foundation in Golden Age Science Fiction. It should not only be possible to have the play list but to tag the play list with the same sort of tags one would tag an individual book with. (This is getting more than a little meta-meta. Stay with me – it’s about to get worse).
As expected, I’d like to be able to rate these play lists. I’d like to see which ones were considered useful and interesting and helpful. I’d also like to see books mapped against each other using the play lists. If two books show on a number of playlists with similar tags then they would be seen as close to each other. For example, without them actually being that similar on the surface, I think that the works of Lois McMaster Bujold and Patrick O’Brian occupy space similar to each other. Am I right? Only voluminous meta-data and careful curration of the retail offering can say.
This arises from the long and noble tradition in science fiction of authors yelling IDIOT at responding to each other’s works. I’d like people to be able to point out the dialogues to me by showing a list of books that take an idea and play around with it.
This is Like… and You Might Like…
To drill down on the play list idea I’d like to be able to associate books or authors with each other proactively. This is a step beyond the “Readers who bought this, also bought this.” I want to be able to say that “I, Robot”, a book about the relationship between homo sapiens and robots, free will and slavery is *like” “Saturn’s Children.” I want to be able to say that the Vorkosigan Saga is like the Aubrey-Maturin books, they both give exposition to their world by having major characters with very different backgrounds and attitudes interacting and having to explain things to each other, in a humorous way.
Another way of working this would be to have people pro-actively suggest books that I liked based on what I’m buying or looking at. If you buy a compendium of Lovecraft short stories 50 readers suggest you might also like the Laundry novels by Charles Stross. If you are browsing I, Robot, lots of readers suggest you might like Caves of Steel, several suggest you might like Saturn’s Children and one idiot suggests you might like Guy de Maupassant.
You could go cross media with this, or in and out of media and I look forward to browsing Apocalypse Now and being pointed in the direction of the Duellists.
Then, I’d like to be able to rate and review these suggestions and have my ability to rate them in turn reviewed.
Credit for Reviews and for other stuff.
I think Amazon’s ability to sell me things depends on how well they curate their collection. If they don’t have the resources to do it themselves and want to pass that job onto the readers then it is only right and proper that the reader is compensated for their time and energy.
I think Amazon should bring in some form of store credit for reviewing books and reviews and suggestions. It doesn’t have to be much. Just enough to get people in the habit of rating and reviewing. I think the credit should be linked to how well thought of the reviews and the reviewers are.
I’d suggest a one penny credit for each star rating. I’d probably cap this in any one 24 hour period and cap it over a month and a year too. Otherwise folk are just going to spend all day giving star-reviews for books they’ve never read. You’d need to do something about bot-reviewing too.
For the first ten written reviews by a reviewer a credit of £1.00. This falls to ten pence per review for the next one hundred. After the first 110 reviews the payment falls to tuppence per review. Unless your reviews are averaging 7 or more stars on a rolling average of the last 50 reviews, in which case you keep getting ten pence.
Each review that gets a helpful vote gets 1 pence for each helpful vote.
Essentially, your first ten reviews get you a free book or two. Your next hundred reviews get you a free book or two and if you prove any good at reviewing you still get the occasional free book.
I’d apply the same credit system to the meta-data. You get some store credit for providing a review of a reviewer or a review. You get some credit for creating a play list, more credit if the play list is well received. If your suggestions that if I liked book X I would also like Book Y result in me buying Book Y and giving it a positive review then you get some extra credit fro
m Amazon.
no subject
It's useful that you can click on a reviewer's name and see all their reviews, but rather less useful that it includes everything, regardless of category. So for example I recently clicked on a name because I liked the book review and thought I might share tastes, only to be shown reviews of a laminator and various presents the person had bought for their family. None of them books.
no subject
My worry is that Amazon aren’t really a bookseller. They are a seller of things that can be posted through a letter box which has branched out into things that can be sold and transported by courier.
So, they don’t much care if I can find the book I didn’t know I wanted because a) they are busy building other empires and b) they don’t care – they see their job as to sell me stuff I know I want. How I decide what I want is largely irrelevant to them.
Which makes them ripe for a counter-disruption by someone who gets curation.
no subject
It's a difficult space because tastes vary so. I adore Bujold and dislike O'Brian very much, and would rapidly lose faith in a system that compared them.
I am supposed to be working so don't really have the time to expand on this, but am wondering whether what you really want is more like a network of forums such as Ravelry.
no subject
(Vampires have too many teeth and too few curves to be remotely sexy IMHO. For me they are filed in the same box as kittens.)
I guess I want a system that is able to work out that I like Bujold *and* O’Brian and that you don’t and then make recommendations appropriately.
I'm happy to chew on some lemons from time to time but what is stopping me from buying anything is the feeling of being swamped.
no subject
I would not be inclined to generalise on the curvature of vampires.
no subject
I think it’s a vast data crunching job. Uploading your library would help – then it could run some statistical analysis and start spotting gaps in your book shelves. Then it can get more sophisticated as it builds up data. I tag O’Brian as nautical romp and you tag him as dislikeable and it starts to learn where your tastes and mine diverge and only offer up suggestions based on areas of mutual overlap.
(At some point it becomes a weakly god-like AI and starts nudging us to read more Dickens and less Stross.)
It requires the customer to want to invest in the retail relationship too.
From a user point of view I’m not sure it needs to be difficult. As output all I’m asking for are a list of ten books I might not have known I wanted a way to tell Amazon it’s way off beam or spot on. If I want the mechanics of the decision it would be helpful to be able to drill into them but not necessary for the output to be useful
no subject
no subject
no subject
no subject
no subject
no subject
no subject
I remember finding it enchanting.
no subject
no subject