The Sunday Salon: Why it's okay to have robots recommend books

Sunday, September 25, 2011


The Sunday 
Salon.com

The question of man vs machine, physical vs online has always been a sensitive one. Whether the topic is why we should support independent bookstores instead of Amazon, why we love a print book more than an ebook, or why recommendations from a person is better than from an algorithm, there are always a lot of strong arguments and complex issues quoted. It's funny that man invented the machine to make his life simpler and he then turns against the machine because the latter is usurping him. Which is to be expected. Heck, I'm going to be severely annoyed if they invent a machine that can write code and design software products, because that's my job. So it's not hard to see the perspective from which the naysayers come. After all, we all have bills to pay and mouths to feed, right?

That said, when Goodreads announced the release of its new Netflix-style Recommendations system, I wasn't expecting much resistance. And true enough, there wasn't any brouhaha. Many, like me, were waiting for it to roll out from the time Goodreads announced its acquisition of Discovereads (the link basically redirects to Goodreads itself), in March 2011. Goodreads is my very favorite bookish site, the place that actually introduced me to book blogging, the site that linked me up with this hidden community of book-ravenous people. So their announcement of their new better-than-Amazon, something-like-Netflix recommendations system earned a big cheer from me.

Still, over the last couple of weeks, I have come across a few articles on the idea of recommendations itself - whether robots or humans are the best sources for picking our next book to read and how machine recommendations are lame when there are living breathing people who can do the same thing, with the added bonus of actually hearing them talk about the book to help us decide if it is the right thing for us. And that's exactly why I read book blogs. Especially the blogs, whose authors review books from the subjective point of view as opposed to the objective point of view. I read to learn more about different kinds of people, about history and its consequences, and also to learn about any topic under the sun, from ancient civilizations to viral diseases. But the reason I keep reading book after book, turning page after page, eagerly devouring the print word is because of the experience of reading itself. So when someone describes how wonderful a book made him/her feel, I am more inclined to read it.

I also listen to Books on the Nightstand podcast, whose 'Two Books I Can't Wait for you to Read' is my absolutely favorite moment of each episode. Some of the recommended books are probably not my usual fare, but the presenters - Michael Kindness and Ann Kingman - do a fabulous job of talking about the books that I still make it a point to read a few passages from the books.

But I love machine recommendations as well. Much as I won't hear the computer talk back to me about how moving or funny a book is, I like it that I don't have to explain my likes and dislikes and issues to someone before hearing him/her respond back with the perfect book for me. I like it that it does all the homework for me and tells me how Leila Aboulela's Lyrics Alley is something I may like because I enjoyed The London Train and The Secret Lives of Baba Segi's Wives. Besides, as Time puts it, "While peer recommendations are important, it's hard to argue against math". Math may have no feelings, but it's accurate. I also love how sometimes I can get choices that I have never heard of, and the element of surprise is retained when I actually pick the book to read, because I know absolutely nothing about it, and to the best of my knowledge none of my favorite bloggers have read it recently, so I feel that I may have a recommendation for them too. To me, peer and algorithmic recommendations go hand in hand, just like you look at Netflix for inspiration when you can't find a friend to talk to about movies.

Of course, Goodreads isn't the first book site to come up with a recommendations system. There have been many before it - What Should I Read Next?Your Next Read and one of the most recent ones, Booklamp. But this is the one I've been most impressed with. The fact that Goodreads' recommendation algorithms not only look at the genre of books we enjoyed but also at user data (what other users thought of a book) probably has a lot to do with that. I tried to get a few recommendations today based on the three Newbery medal winners that I've already read so far. And out of the 24 recommendations the site threw at me, only three were titles I wasn't interested in, and even then only because they were books of poems. When I checked the literary fiction folder, there were a ton that has me eager to hit the library right now. There's something wonderful about seeing books you have probably never heard about, and that's what gets me excited about seeing mass online recommendations.

Do you like getting recommendations from a non-person?

37 comments:

Cat said...

Overall, not really. It works sometimes but all too often I just get books I've already read or I end up not liking. Much prefer recommendations from real people who I know share my taste in books.
You may like Lyrics Alley - I didn't!

Have a great reading week!

Juju at Tales of Whimsy.... said...

While I don't mind them, I find the best recommendations come from bloggers with like taste as mine. Great question/topic.

Harvee said...

Thanks for pointing out Goodreads' recommendations system. I tried it and I can't wait to read some of the books it suggested for me based on my reading! Booklamp had only a few of the books I've read so it's not as useful for me. It seems to have only the best sellers. The recommendation system beats browsing the library hoping to catch a good book, though I've found many that way, as well as through recommendations from bloggers.

Athira / Aths said...

So far I haven't got any recommendations from Goodreads that I have read but that's because I have almost all the books I've read cataloged there. I am yet to read one of those recommendations so I'll know better about whether it works for me or not. That said, I had tried reading robot recommendations from other sites, and that didn't always work well. I'm hoping that the different way in which Goodreads recommends (they use user data itself as opposed to book characteristics alone) will help in turning up great books.

Athira / Aths said...

I agree with you there - I am also more inclined to pick up a book recommended by a blogger vs a machine. The one time where I like robot recommendations is when I want to pick up a book without knowing much about it, or in genres that aren't well read, like graphic novels.

Athira / Aths said...

I tried Booklamp but I wasn't impressed either. I rarely read popular books, so they didn't have what I was looking for. I'm hoping the Goodreads recommendations continues to give good options. 

Kerry M said...

I've been using Goodreads for a while now to keep track of what I've read, but didn't realize that they had rolled out a recommendations feature. Librarything had something similar, but I find Librarything a bit too cumbersome in terms of adding books that I've liked, so the algorithm was always a bit skewed.

I like your thoughts on robots vs. people recommendations. While I, too, am more interested in picking up books because someone else told me a powerful story of how much they enjoyed the book, there are certain times when I really just want to know what to read next and I'm not much up for gambling. This sounds perfect for that!

Colleen said...

Great topic! I think there is a place for both. Working in a data driven field has left me with a bias towards hard data which I see bleeding into other areas of my life so the algorithm for book recs appeals. But I also appreciate recommendations from friends and bloggers.

Athira / Aths said...

I did try LibraryThing as well and found it messy, so I had to kiss the site goodbye. I do like Goodreads a lot - it works well for all kinds of bookkeeping I do. I like machine recommendations primarily because they have the gambling factor in them. Sometimes I like not knowing much about a book.

Athira / Aths said...

I'm the same way - I like the reliability of math data (if it's based on a well-charted algorithm), and that's what makes me want intelligent recommendations systems (based both on the type of book and on what other users thought of it).

ChewDigest said...

Dare I say that I prefer robotic recommendations? There is less pressure and, in the case of GoodReads, the robot is looking at everything I have read and rated as opposed to the few that the recommendee might know I liked. 

My only issue so far with the GR choices have been in the folders that I have for decades. Whether fiction or nonfiction, if it took place in say, the 1910's, I put it in that folder. GR came up with a bunch of stuff that while similar to some of the subjects, had nothing to do with the 1910's. However, there is no way that I would expect a computer to realize why I had grouped them the way I did, so it is forgiven. 

The other bonus is that a robotic recommendation doesn't come with all of that pressure. If you tell me that I "just have to read this book!" and I hate it, I feel bad. If a robot tells me that I might like something and I don't...I feel fine:)

I just tried booklamp and that bombed. There is a new one that is still building up a database and makes recomendations from a slightly different angle, customreads.com. I am still playing with that one. 

Vasilly said...

I love both types of recommendations. I haven't tried the latest feature from Goodreads yet but I know I will soon.  Great post. 

Athira / Aths said...

Thanks! I hope you have fun with the recommendations feature.

Athira / Aths said...

"There is less pressure and, in the case of GoodReads, the robot is looking at everything I have read and rated as opposed to the few that the recommendee might know I liked." That is primarily the reason why i like robotic recommendations. I can choose it or not, and no one feels bad. Also if I choose to read it, I can happily say that I hated the book. I have to check out customreadds! Thanks for sharing!

zibilee said...

I think I prefer getting my recommendations from a real person because by reading their opinions I tend to get a more full impression of the book that is being discussed, but I am not averse to trying a robotic recommendation, and have not checked that out yet. It does sound interesting, and kind of cool, but I don't think it will ever totally replace reader recommendations for me. This was a very thoughtful and interesting post today, Aths, and it has made me really curious to try out the new feature at Good Reads!

Erin said...

I'm with you on why I read book blogs. I like reading about personal reactions to books. Oftentimes a blogger will highlight something I particularly dislike in books, or vice versa, that will make me decide to pass on a book I may otherwise have picked up...or vice versa. If that makes sense!

I haven't really messed much with robotic recommenders. I like to peruse the lists they generate but haven't yet taken their suggestions under serious consideration. I think I'd like to at some point, though, just to see if they work!

Lisa said...

I'm all for recommendations from any place I can get them. Goodreads knows everything I read and can compare it to a huge number of other people to see if it can come up with books I might like. But I when I checked out the Goodreads recommendations, I wasn't overly impressed. But I'll keep an open mind - because you can never get to many books on the wish list!

Helen Murdoch said...

Machine recommendations freak me out a little, especially when they are so accurate and spot on!

Zohar said...

I like the Netflix recommendation system, but to be honest the first thing I look at is the stars on Netflix. I know that nine out of 10 times I'd like anything that has four stars or higher.

However, that is not a machine recommendation system is it?

When I read recommendations from blogs I usually put more weight on bloggers who have my taste. Reviews (of anything, would it be books, restaurants, movies, etc.) are all individuals - the trick is to find someone who agrees with you most of the time.

http://www.ManOfLaBook.com

Care said...

I have yet to see what goodreads thinks I should read...    but I am curious.   

Giving Reading said...

I think I totally agree with what you said here :) 

LouiseReviews said...

I've just joined up to Goodreads - so not sure how it will work for me.  I tried Book Lamp but it didn't have some of my favourite books on record (To Kill A Mockingbird for example).  I'm still trying to get round to all the books that real people are recommending but I'm always willing to give things a go....

Kim @ Sophisticated Dorkiness said...

I just played with the GoodReads recommending tool a bit, and it seemed like it was pretty accurate after I cleared up the genres I was interested in. I'm not sure I'd take my recommendations exclusively from robots, but I think it could be good for showing older books or ones not getting buzz from the bigger book recommendation places I go to. 

Kalpana Palaniswamy said...

Well, I may not be comparing apples to apples here. I saw a TED video yesterday by Abraham Verghese, author of Cutting for stone. http://www.ted.com/talks/abraham_verghese_a_doctor_s_touch.html 
I have heard about his book, but after listening to this TED video, I have bumped its order in my TBR queue. I value human touch more than anything else. I would like to use the computer only as an aid. 

Alex (The Sleepless Reader) said...

Book bloggers are a great source of recommendations once you start realizing who likes the same things you do. I also love to check the "favorites" shelf of my friends on GoodReads.

I'm willing to give GR's recommendations a fair shot. Only browsed a bit and already added a few to my "to-investigate" list. in the end, wherever the recommendations come from, I'm always open to new ideas!

c b said...

I have a list of books I abandoned on goodreads. I should see what GR recommends based on that category, and NOT read them :) I welcome both human and non-human recommendation - I just love hearing about new books, then it's up to me to decide whether I will read it or not... I don't just read everything that's recommended to me.

Athira / Aths said...

I look for different things from robot recommendations. Mostly, I go to it when I want a surprise choice, a book I know nothing about. 

Athira / Aths said...

That's just my approach too. Sometimes I may not wishlist a robot recommendation because it doesn't sound that great, but then a blogger will talk about all its fabulous aspects and then I know I just have to read it.

Athira / Aths said...

And that's why I like goodreads recommendation. The site already knows my reading tastes well. And then it compares my tastes with that of other readers, and how wrong can the site be, especially if you already have a ton of books listed?

Athira / Aths said...

If you had a chance to check it out, I will be curious to know if you agree with it or not. 

Athira / Aths said...

Thanks!

Athira / Aths said...

I doubt even the Goodreads recommendation is a pure-machine thing, especially if they are depending on user ratings and not just on book qualities to suggest reads. But mostly, I just like the serendipity aspect of the whole online reco thing.

Athira / Aths said...

I did check out Book Lamp too, and had the same problem - it didn't have a few books that I loved. I hope you like Goodreads though!

Athira / Aths said...

That's my ploy too - plus if I want a few recommendations of books that I have never heard. 

Athira / Aths said...

That's all I do too. Most robotic recommendations I've come across before haven't exactly intrigued me. The Goodreads one is actually pretty much reliable, because it did seem to recommend me books based on what I enjoyed and what people with similar tastes enjoyed.

Athira / Aths said...

I love getting recommendations from book bloggers too. But when I want to pick something that I have no clue at all about, I like to go to machine recommendations.

Athira / Aths said...

Haha! That would be a fun way to not pick books! I will also have to try that. :)