Can’t we all relate to this feeling, arriving on the Netflix homepage and being overwhelmed by the number of movies to watch and tv-shows to binge? Ever indecisive due to all the options we are presented with that might be more entertaining, educational, or somehow worthwhile. Overwhelming to the point where we would be genuinely stuck for 20 minutes whilst cycling through the queue trying to decide…
This feeling, however, is not exclusive to Netflix. When choosing something to read, watch, listen to or do, we have endless possibilities and seemingly meager tools or services to sort through our options. The ability to connect with what we truly want to discover and interact with has become more challenging yet ever more important due to the number of options we are constantly facing.
Surely we can ask friends for suggestions, we can review aggregated ratings, and we can browse domain-specific platforms, but too rarely can we find the best possible options, uniquely suited to us at a specific moment. This is specifically why we need deeply personalized recommendation systems.
The number of options we have access to nowadays is exponentially growing: millions of songs are available on Spotify, thousands of shows and movies streamable online, hundreds of restaurants… Giving need to “information filtering systems”, and more specifically recommendations systems.
Recommendations are so common that it becomes easy to overlook the seamless way recommendation systems have been incorporated into practically every device and platform. A recommendation system is defined on wikipedia as ‘a subclass of an “information filtering system” which aims to predict the rating or preference a user would give to a specific item’. More than that, recommendation systems are “ discovery engines”. Tools to assist us by optimizing our personal choices, matching our individual preferences across a large body of work, seeking out the perfect personalized match.
But recommendations systems are not needed solely to find a ‘needle in a haystack’. The reality is that, throughout our entire search process for the “perfect item” we need recommendations in order to be less anxious, to have the luxury not just specifically to find an item, but to actually ‘look for’ it. This confusion between ‘finding’ and ‘looking for’ can be clarified in the sense that ‘finding’ what we do on our own, the final moment when one makes up is mind. We rarely received one recommendation spot-on, we receive a curated list of suggestions, and then we decide.
The reason why I talk about being less anxious or being happier while searching for an item is because this abundance of propositions and items triggers a real distress. This trigger was called “the Paradox of Choice”. First introduced in 2004 in the book The Paradox of Choice — Why More Is Less, American psychologist Barry Schwartz defends that an overabundance of choices first paralyzed the consumers, but more importantly, creates anxiety and frustration.
There are now several books and magazines devoted to what is called the “voluntary simplicity” movement. Its core idea is that we have too many choices, too many decisions, too little time to do what is really important. […] Taking care of our own “wants” and focusing on what we “want” to do does not strike me as a solution to the problem of too much choice. — Barry Schwartz
This sends us back to our “Netflix Homepage” situation. People are often paralyzed by choices and need more individually designed recommendations. Another factor is the fear of making a “sub-optimal choice”. The FOMO (Fear Of Missing Out) the right items creates in addition a deep frustration and un-satisfaction. A concrete illustration of these phenomenons is given by those simple metrics: the average worldwide Netflix user spent 21 minutes a day or 126 hours a year looking for something to watch.
This is why personalized recommender systems are meaningful in the first place. By filtering in this abundance of items the ones that would be relevant to one user, it allows in the first place to eliminate the noise, to reduce the anxiety, to reduce the frustration, and to really give us the opportunity to decide among a relevant set of propositions.
But if recommendations are presented to me, aren’t they already tailored to my unique set of tastes and preferences? Why am I insisting on the “personalized” aspect?
To answer that, we would need to consider a more mathematical approach of recommendations systems.
Typically, the most used mathematical approach for recommenders system are “User-Item Matrix Factorization”, or more commonly called “Collaborative Filtering” or “Social Filtering”. Collaborative filtering filters information by using the tastes of other people. It is based on the idea that people who agreed in their evaluations of certain items in the past are likely to agree again in the future. For example a person who wants to see a movie might ask for recommendations from friends. The recommendations of some friends who have similar interests are trusted more than recommendations from others. This can be summarized by the sentence “people who have the same taste as you liked this, so you will like it too”.
But the overall consequence of abusing this approach (like the majority of businesses) has more impact than we think. Individuals are mathematically represented as a group, and groups as masses. A series of clusterization that progressively leads algorithms to deliver recommendations based on how similar we are to other users, and not on what makes us unique.
One compelling illustration of this clusterization is found in the case of Netflix, where “more than 80 per cent of the TV shows people watch on Netflix are discovered through the platform’s recommendation system”. In their specific case, recommendations are the results of cohorting users into ~2000 different taste groups. In other words, for Netflix’s total subscriber base (approximatively 250 million active profiles) this would mean that one user is represented similarly to 250m/2,000 ~125,000 others individuals..
Another consideration of our uniqueness is the context. It’s all relative, isn’t it? The easiest illustration would be to consider yourself, and see how contextual changes affect your decisions and desires. For instance, what music do you listen to in the morning and what music do you listen to at night? Are they the same, or do time and events of the day influence your decision? Mood, weather, temperature, sleep, emotions etc... All these factors are key elements in understanding and predicting what you might want or need at a certain moment. However, these are still never fully considered or even acknowledged by businesses and media platforms.
We have now become algorithmically represented based on the similarities we shared with our neighbor in addition to being deprived of our context… Where is the uniqueness in this?
How can one service claim to find the best movie for me tonight if I know there is no way it encompassingly understands who I am, at this moment?
The need to demand “personalized recommendation” becomes equivalent to the need for valuing and acknowledging the uniqueness that we have as human beings, individuals. Because without it, vicious cycles arise from these interactions, creating problems like “filter bubbles” that hinder self-development and exploration. Many recommendation engines operate in moral hazard zones through the Principal-Agent Problem: concerned with profits and addictiveness, they don’t have user’s long-term, best interest at heart. Therefore, by leaving this fight, being at peace with the mediocrity of the recommendation platforms are delivering to us on a daily basis, we implicitly give our approval. We agree that “what you [media services] are creating and presenting to me will satisfy my binging addiction for a moment”. The goal of those platforms is to create and promote something that will satisfy the mass, not ourself. And by not asking for better, by not reclaiming our identity, we implicitly agree to be seen as an element of this mass, and not as an individual.
It is with all those considerations at heart that Dr. Emile Contal and myself co-founded Crossing Minds. We have spent our careers seeking to develop contextually-relevant, optimized recommender systems to deliver the best possible personalized recommendations for users at the most opportune time. Through user-filtering, item-filtering, content-filtering, deep learning, and transfer learning, the recommendation systems developed by the company seek to surface the best recommendations unique to each user, not simply for “users like you”. That’s the company promise.
We are currently open for sign-ups to our Early Access program for hai, our first experience focused on delivering the best recommendations for media you’ll love but haven’t discovered yet.
To check it out, you can sign-up here.
Thank you for reading!