The contents of the web are endless and it is this very nature of the web that makes it a great resource and an uncharted jungle, all at the same time. The problem the web now poses is how to get to what you want and how to get there fast.

While content consumption patterns on the web are wide and varied, if you were to plot them on a graph, the pattern that would emerge would have  these at the ends of the spectrum  –

At one end, you will have channels that you can subscribe to, like twitter. The premise is simple - just follow people who tweet what matters to you, and enjoy reading.  Sounds fairly simple, but in practice, the stream is likely to become overwhelming very soon. You have a lot of content and choosing what is important becomes a challenge in itself. Added to this, as the number of articles increase, the quality of content begins to drop too. This is a problem that plagues most broadcast systems.

At the other end of this spectrum would be your haven if you are a control maverick. You go to specific sources of your choice and read articles that interest you or query on google for what you want and read what shows up. While this ensures quality and control, the time and effort needed to get to relevant articles becomes a limiting factor and you end up reading fewer articles. You miss out on articles that are important or interesting to you.

One perspective to Content Discovery is to say that it occupies everything that lies between these two ends of the spectrum. It is a genuine effort to bring you as much high quality content as you wish to consume, but let you stay in complete control all the time. There are quite a few systems aiming to attain this perfect balance and we are in contention too.

With a plethora of Content Discovery Tools (CDT) out there, what should we look for in the good ones ? Here is a short list ..

1) Continuous learning  -  Any CDT needs to learn about you to serve you better. What differentiates the good ones from the also ran’s is how much should they be told about you and how fast they learn from it. The learning could be explicit - you tell the system that you like a particular article by giving it a thumbs up, or you say you want more from a given author or source. Learning could also be implicit, by mining into your reading patterns, pages you tend to visit or read often or the kind of articles you prefer sharing with your peer group.

2) Recommend with reason  -  A CDT should give you proof of why something has been recommended for you. This is important for two reasons :

a.  It makes you aware of what the system has learnt about you

b. It gives you an opportunity to fine tune that learning

For E.g. : If the system recommends an article, a likely proof could be a similar article you have told the system you like.

3) Context  - Context is what makes or breaks a recommendation.  Put simply, context is what you are doing now. A highly evolved CDT gets the context right and this enables it to compute what you are most likely to do next and aids you in that effort. It might even lead you there. Here is an example..

To summarize, a good system learns fast and recommends with reason in the right context.  We have gone beyond the above list and we wish to move the needle higher by bringing in more dimensions into the content discovery experience  -

1) Social endorsement - We believe that content discovery should have a social angle to it. Here’s why :

a. In a survey we did very recently, close to 70 % of the respondents said that their colleagues either visit the same sites or read on the same topics as they do. This means, there is a lot to be gained by hunting in packs. Your peers do not have to reinvent the wheel when it comes to finding great articles to read. We have a one click approach to make content socially discoverable in the right contexts.

b. The trust you place in a person who works in our team or in your close friend is usually an order of magnitude higher than the trust you place in a publisher, even a well know one at that. So, recommendations from your immediate network are positively influence your consumption.

2) Anywhere, anytime - It is a given that you have multiple interests and these interests lead you to different sources on the web. A CDT should compliment your preferred sources with relevant articles from new sources worthy of discovery. It should never restrict your consumption to a particular site or a small set of sites. It should be an indispensable part of your web journey and provide you with great stuff no matter where you are.

We will soon be launching Pugmarks – a product where we will bring all our thoughts on content discovery to life. Do register for an invite on Pugmarks.me and stay hooked!

Aditya