Why do I feel bad following recommendation algorithms?
How can I still find interesting things without them?
When I was younger I used to roam the library for a new book to read. I would look at the cover art and read the back of the book to see if I thought the story was worth reading. Sometimes I would just find another book written by an author I liked before to make sure I found something interesting. This process was enjoyable and resulted in some great finds, but some terrible picks as well.
Times have changed. And this isn’t some recent thing either. On the subject of literature there is Goodreads1, which has been providing recommendations to users since 2011. But the rise of recommendation algorithms evidently has not been confined to the realm of literature. In fact, most of the media we consume has some sort of recommendation algorithm that provides you with suggestions based on your past behavior and preferences. The result is a tailored experience in which you only see content that is interesting for you, reducing the risk of choosing things not relevant for you.
In recent years I have been struggling with recommenders in general. I would just watch hours of YouTube content in a row and there was always something relevant for me to watch. Whenever I couldn’t focus on something, I would open up YouTube and find something to watch. This would be such a timesink. It was no longer a deliberate choice to watch content I wanted to watch, I just wanted distractions and the algorithm always found something that looked “interesting” to me.
But what is the difference between following recommendations done by an algorithm or done by a person? Reviews for any media have been quite influential for a long time and in a sense they don’t differ much from the recommendations by an algorithm. On the one hand the reviewer notes their arguments for their review of the book, but they are not aware of your personal tastes. However, the algorithms will provide you with tailored recommendations that you’ll most find enjoyable, but won’t give you any insight in their reasoning for it.
Yet I still prefer buying a book based on it’s reviews or recommendations by others rather over the recommendation by an algorithm. When researching this further I found a publication in which professor Carey K. Morewedge2 notes the fundamental constraint of algorithms is that they only learn from our behavior, rather than our actual preferences. On this topic he mentions that:
“Still, the revealed preferences of individual users are an incomplete and, at times, misleading measure of the “normative preferences” that compose their true goals and values.”3
Which is interesting, but not the reason why I’m skeptical about recommendation algorithms. However, later in the article he notes that:
“Recommendation systems help us find what we like, but they also change our preferences and reduce the diversity of what we see and buy, increasing the market share of popular and existing options.”3
This seems to be my main gripe with recommendation algorithms. I want to read something and reflect on it. What did I think of it? Was it interesting to me? Was the story captivating? What specifically did I like? Was it well written? Etc. Just clicking “like” on a book does not let you reflect, it answers these questions for you. And within it’s black box it weighs the criteria itself.
I want to discover my taste profile myself, I want to understand why I like certain things. This will allow me to convey my preferences without getting a label assigned.