Some choice good – excessive choice bad. That is the (condensed) Paradox of Choice, according to Barry Schwartz. We need some choice in order to exercise our free will, but the abundance of options we’re facing today (when shopping for groceries, entertainment, education and more) is actually quite overwhelming and paralyzing. No matter how much research we do, no matter how carefully we consider our options – we always risk feeling slightly disappointed and unsure if we could have made a better choice.
A good recommender will help us decide by dramatically reduce the number of options we have to explore, into a small subset of what is available and potentially relevant. A better recommender will cut the crap, and just present the options we need to make a choice we’re happy with. Nothing more, nothing less. Because too many recommendations can create a second problem much like the initial Paradox of Choice, especially when we our goal is to reach a (possibly very important) decision.
Exploration is often not a goal in itself. It’s a painstaking process we have to endure, in order to feel confident enough to make a choice. As an example, if I want to learn more about user experience design, I can go to Amazon and shop for books on the topic. A quick search returns a list of 1,089 books for me to choose from, and for each product page there’s several more recommendations, user reviews and reading lists I can base my decision on. Amazon is not giving me the answers I need, just endless opportunities for exploration. I feel my blood pressure rising.
In a world with perfect recommendation technology, Amazon would just tell me which three books I should read to learn more about user experience design. From all the 1,089 available options, Amazon would recommend me three well-written and thoughtful books, covering most bases of the topic. Perhaps I get to choose between a set of cheaper books and a set of more expensive books, depending on how much money I’m planning to invest in my knowledge upgrade. In any case, I’ll be happily making a simple choice, free from the torments of exploration. It doesn’t matter if my choice is sub-optimal (there may not even be a single optimal option for me). I’m ready to satisfice, just as long as the recommendations are sufficiently relevant.
Current technology is not capable of giving perfect recommendations (partly due to ambiguous user queries and changing user preferences), so we cover up by design systems that encourage exploration and discovery. We strive for transparency and diversity, designing an experience for the eager and motivated user. But for some of us, the Paradox of Choice returns and takes it’s terrible revenge, as our options for exploration grow and overwhelm us again. Somehow I feel that efficiency (making faster decisions) and satisfaction are better design goals for many recommendation systems.
Exploring music is, on the other hand, something I enjoy doing. My music preferences are strongly connected to my personal identity, and that changes somehow my attitude towards choice and exploration. Choosing music becomes a meaningful act of self-determination, and music exploration becomes a goal in itself. So I cherrish the many options for exploration I’m given by Last.fm.
Kirsten Swearingen and Rashmi Sinha gives us a neat breakdown of different user needs a recommender should accommodate. Here is their list of recommendation types:
- Reminder recommendations, mostly from within genre (“I was planning to read this anyway, it’s my typical kind of item”)
- “More like this” recommendations, from within genre, similar to a particular item (“I am in the mood for a movie similar to GoodFellas”)
- New items, within a particular genre, just released, that they / their friends do not know about
- “Broaden my horizon” recommendations (might be from other genres)
I feel that my music exploration needs are well represented in this list, but I wish for a stronger focus on efficient and satisfactory decision-making.