Book Discovery has had a terrific last few weeks. Clearly the highlight was the ‘it’ app / tech product discovery site Product Hunt’s expanding to books.
Just before Product Hunt Books launched, Jordan Koschei launched fivefootshelf, a website devoted to curating essential reads by topic. And before that there was the Bookstck v Bookicious war (each curating entrepreneur-recommended books) on Product Hunt.
Clearly book discovery is seeing a lot of action. Why?
I like to think of book discovery as a subset of discovery problems in general. Thus far we have been using search engines to find what we know we want. And search (Google Search, really) is a terrific tool for doing this. But what about finding or discovering what we didn’t know we wanted. How do you search for something you dont know? This is the challenge facing text search as the dominant paradigm for discovery.
To add to the above, we are witnessing Google gradually losing its indexing land mass – more and more time is spent on sites (Facebook, Pinterest) or on apps (mobile, or desktop apps such as Slack / Quip) where googlebots are banned, affecting Google’s ability to index the site. This affect search quality, which further heightens the discovery problem.
Silicon Valley hasn’t been sitting still. Pinterest, Siri (or Cortana or Google Now) etc are all attempts to take on discovery head-on. However like all grand challenges, discovery isn’t going to yield with one particular line of assault. It will need attacks from multiple directions. And this is where book discovery comes in. Book discovery is a great training ground – akin to what the playing fields of Eton did for Waterloo – for that grand assault to come. Further, there are advantages to books as a category too – books are a reasonably finite closed set, there is lots of meta data, and they attract a lot of passionate readers, who can always be counted on for feedback.
Book Discovery thus far
Up until now, book discovery sites have used either crowd-curation or code-curation to attack discovery. Crowd-curation uses social book shelving (a la Goodreads, the ur-company of this era) to attack book discovery. Typically such sites are free to users, who add books to their shelves, comment on the same, follow their friends and see what they are reading to determine what they should read.
In code-curated sites, algorithms are used to determine what you should read next, basis your past history and preferences. Typically these sites involve less friction (and set-up time than social book shelving sites) but also result in unwieldy recommendations (as you see with Amazon recommends).
The rise of expert-led curation
What scale giveth, relevance taketh away. Given the tradeoff between scale and relevance, it was just a matter of time before the lack of relevance pushed entrepreneurs to look at new routes. And that is what the emergence of sites such as Bookstck / Bookicious and even Pinax tell us. Each of these uses expert-curation (or human-curation as Mark Watkins refers to in a series of LinkedIn posts).
Interestingly, Product Hunt for Books is a twist on expert-led curation itself. In the case of PH for Books, the community is highly curated, even though the books aren’t. Through discussions, there is scope for recommendations to emerge, but it is also likely that you will see books that aren’t recommended by any expert. The books’ presence in the curated community is in itself the recommendation!
As these sites gather steam, increasingly they will have to move beyond the present stage of manual aggregation of expert recommendations, where someone needs to physically collect recommendations, or ask someone trusted or knowledgeable to recommend. The challenge here is really the relevance of the expert to the reader. Sure I get to know the very best books that Marc Andreessen has recommended, but if I don’t know who he is, what impact does the curation have? Thus relevance of subjects / experts become key to maintaining the experience.
To meet this, and simultaneously bring in some degree of scale, we will begin to see algorithm-led aggregation of expert recommendations – here you could use some sort of algorithm or automation to collect recommendations, but more critically, crunch the recommendations of the people you follow on twitter (as a proxy for relevance), to come up with a definitive list (Bookvibe comes closest to this but is still miles away) of books that you are bound to like. This is what the present bookstck, fivefootshelf or pinax have to move to, and this is where the future of book discovery is leading to – machine-driven expert (human) curation.
As I conclude, it might also be interesting to keep in mind that the recently launched Apple Music is also a clear bet on curation through their curated playlists feature. Music is actually the category that has maximum metadata – Pandora classifies a song under as many as 450 attributes including moods, making it easy for its algorithms to parse your present playlists to suggest the next song. Still Apple Music thinks the future is human curation. Fascinating!