Faced with the overwhelming amount of online content, choice becomes difficult. What to read and from where? All too often, the almighty fear of missing out prevails. Indeed, we feel inundated by an avalanche of content, and although this may be the way it is, it sure doesn’t feel very good.

The filters that naturally tend to compensate for this phenomenon – just like in real life – are those based on the behaviour of our peers: “Tell me what you’re doing and I’ll know what to do.” The strength of social media is that it allows you to direct your decisions or actions based on what others are doing or saying – like before, like always, but at a much stronger level. Nothing really new here, except that its prompted a few interesting ‘social filter’ initiatives popping up as a result.

Here are a few such initiatives to prove my point: Stamp is an example of a social filter for cultural addresses or tastes. Dunno is an example of a social filter for information searches that we can share and build upon. Wavii filters by thematic content, compiling information sources according to your friend’s traffic levels. And let’s not forget to mention Netflix, Spotify and etcetera, all of which allow me to see and know what others are watching.

BUT… These filters remain the reflection of others – and not necessarily a reflection of ME. Who are these other people? By depending on others to build my own Internet reference universe, I don’t necessarily get the best input, and, to my detriment, end up being influenced by the tastes and lives of others. Not to mention there’s the awkwardness of discovering the hidden passions of certain friends, the boredom that comes with the same old themes or subjects that aren’t necessarily of interest anymore, and the wasted recommendations that just don’t suit me at all.

Herein stems the resurgence of models focussed on the individual. This time, it’s not an old-fashioned, “self-service” type of personalization (choose your sources, how to organize the page, screen colour…remember Netvives or iGoogle)? Three major models stand out from the lot, with a mission to learn my tastes or my habits.

To me, the predictive approach is the most fascinating, with Google Now and Saga being the most emblematic of the lot. They are based on the analysis of data obtained from the study of online user behaviour to provide information they need, when they need it. Predictive customization is, for example, getting the status of the Bixi station closest to your home on your cell phone at 5:25 because the application knows that you will need a bike at 5:30; it is about getting the metro status 10 minutes before leaving your house because this is how long it takes to get to the station. Overall, this approach involves the ability to predict the moment someone will need information, not only because it’s of interest, but because it also corresponds to thier lifestyle. On the other hand, the user should not be afraid to share their personal data. If the best possible service is expected, then the user should provide the same in return with regards to data sharing.

The learning approach is best represented by the Zite application. It shows how a site or application learns from the interests and behaviours of the user. Zite evaluates the content provided, analyzes whether or not there is any other similar content, and in return offers content that increasingly reflects the user’s needs. Zite consists of an exchange of data with the user, but it will only work if it is authorized to analyze browsing history, location, etc.

As for the selective approach, it simply allows the user to save content to read later, forming a customized magazine based on the information selected. Instapaper and Pocket are key versions of this approach, acting as a “smart” bookmark. The selective process allows the user to select the content he/she wants without having to provide data.

A special note: we could discuss the Google Now model at much greater length. Today, it is getting a head start on sketching out how we will access information in the future with devices like Project Glass (the famous augmented reality eyeglass prototype unveiled by Google a few months ago), or even intelligent TV. Because I won’t be able to easily interact with a big screen, glasses or what have you, it’ll be imperative that the device knows what information should appear at what time, without me having to ask for it…

Obviously, these predictive or learning personalization models don’t come without any strings attached.

On the one hand, they emphasize the transformation of an audience that leans more and more towards specialization, but is less inclined to discover by surprise. If I’m not given what I like to see, how can you recreate the taste of discovery? The element of surprise? The chance to learn something new? (We don’t start out liking wine or cheese, it’s an acquired taste. Like culture. Like new things or experiences). Editorial experts will most likely get to tackle this challenge.

On the other hand, they involve a new connection with private life – that of a public private life, with open data. It’s a radical trade-off, because the only way to let technology become an extension of ourselves (and even capable of predicting our actions and needs) is to accept sharing data like never before. It’s up to each of the stakeholders (media, brands, readers, consumers) to establish a close, new relationship between data and confidentiality.