The Discoverability Problem: How To Get Out of the Filter Bubble Recommendation Systems? | KnowledgeManagement |


IdeaEncore:  Discovering what people 'like me' are doing is one way to discover new content. But, it all comes down to how we define "like me".   For a professional knowledge sharing site, like IdeaEncore, it might be topic of interest or sub-sector or functional role or experience level.  But the most interesting, breakthroughs happen when we *cross* these traditional boundaries and explore solutions in adjacent or even completely different realms.  So maybe the 'like me' algorithms don't have it right.  Maybe the 'trick' to to discover by 'problems like mine' 



Robin Good: Brett Sandusky attacks the "discovery" topic with simple, straight logic, analyzing what all the new startups and the new tech fanatics seem to systematically look over.


How can you help me discover new stuff, if you are intentionally limiting your exploratory gathering to algorithms and to, however varied, network of contacts?


She writes: "The discoverability problem in books is a challenge. It’s about connecting users to new and interesting titles, that they wouldn’t normally have seen. This last part bears repeating: …that they wouldn’t normally have seen.


Ultimately, the problem with all these discoverability sites is this: their algorithms (if they are even using an algorithm) are based on aggregate data in a one size fits all model.


The more people who read something, the more often it shows up in your recommendations. But, that’s not discoverability. That’s the NYT bestseller list. That’s Nielsen Bookscan telling you the top sales of the week.


Just because most of my friends are reading bestsellers (because, duh, whose aren’t? In fact, that seems to just reinforce the concept of the term “bestseller”) does that mean I should only be shown these titles?


Obviously, the answer is no. But, how do we get there?"


The answer is that we need a) more expert and qualified human intervention to unearth and pick new stuff, and b) behavioral data coupled with data collected on customer preference to allows us to connect those selected materials to the users in the system.



Rightful. Timely. 8/10


Find out:


(Image credit: Josephine Wall - Discovery)



Via Robin Good, IdeaEncore