Discovery has become one of our main area of interest at Diveboard. Of course discovery isn’t a new idea – helping someone stumble onto something that will get him excited has been on the mind of every marketer forever.
What has changed greatly though is the means we can leverage on to get there.
Discovery actually dates back from the 50s as a part of Cybernetics. For those like me who were born after that discipline stopped being taught, we may have a wrong idea about what cybernetics is. Cybernetics has little to do with bionic lungs and enhanced humans, its actually defined by wikipedia as :
Cybernetics is a transdisciplinary approach for exploring regulatory systems, their structures, constraints, and possibilities. (…) Concepts studied by cyberneticists (or, as some prefer, cyberneticians) include, but are not limited to: learning, cognition, adaptation, social control, emergence, communication, efficiency, efficacy, and connectivity
This basically relies on the concept of the feedback loop which enables a system to adjust to its environment. Here’s a simple example of such loops:
Internet really exploded when proper search got in. The basic idea of search is simple: you need to be able to express what you are seeking and the algorithm will try to “understand” it and find in its database a match. Basic search is easy, but as Google and others have demonstrated over the years understanding a query properly can be a huge help in finding the most relevent information – this is called as the Semantic Web.
This really makes a lot of difference when it comes to giving you the answer vs finding a relevant page.
For example let’s say you’re looking for the best New York marathon time.
wolfram’s answer versus google‘s are making that difference very clear.
But as search hits a hard limit – since there’s as much you can understand from a query – or at least as the cost of getting a little bit better has raised exponentially, the whole discovery space is getting momentum.
Discovery tries to read the user’s mind, understand it’s behavior, learn from it and try to get to the user contents related to its research but without limiting itself to the keywords he may have typed. Discovery is about digging relevant stuff in a given space.
The first notable forays where done by such like Amazon using what is now called “Collaborative filtering” techniques. It’s basically a “people who liked this also liked those”-type of algorithm. This happened at the turn of the century and this simple methods have given spectacular results.
Today, as we learn to aggregate manage big data, new doors are opening up. Numerous research projects, and some implementations are starting to bring back those old ideas from the cybernetics age to build pesonalized feedback loops for each of us.
By testing us with stimuli and watching us react (click/scroll…) the system can learn a lot about our affinity to a subject and hence profile us and serve us with more relevant content that will be appealing to us.
In the very specific case of Diveboard, that would mean understanding where a scuba divers wants to go next (and why) and what he expects to be doing there : is he more excited about big fish, caves… does he start considering taking a training course to level up … all those things are already in the system and would bring a lot of value by directly pointing a user to the elements that matter to him instead of flooding him in an endless listing of stuff he wouldn’t care about or -worse- forcing him into formalizing what he’s searching for, provided he’d know that himself.
Since we’re spending a lot of time on this subject I’ll follow up on this note on some analysis we’ve done about how Discovery is currenlty implemented on the web and how it’s performing.