When I'm out of town and looking for somewhere to eat I open maps and pick somewhere near by with enough stars to lessen fears of food poisoning. These ratings come from humans that have been there and wanted to voice their opinion. Ignoring selection bias, the star rating is a good way to avoid terrible food but not the path to finding the best. Whether it's food, films or fiction, public ratings are everywhere. As more companies try to convince us to consume more, these ratings help dictate what you should check out next. AI knows what you've consumed, finds others with similar taste and shows you what they did next. When recommendations mean money, avoiding terrible recommendations is a good metric to work on. When we optimise our own lives however, we need to remember that sometimes you need to kiss a few frogs.
We've never had quite so much to choose between. There are more books, movies and meals than one could feasibly consume in a lifetime. Given that we rely on recommendations to guide our choices. Long ago that may have been reviews in the paper or from a friend of a friend. Today, those recommendations are mainly delivered by AI. AI aiming to personalise recommendations to your unique taste. These algorithms optimise for hours watched or purchases made and tend to follow a pattern of mimicry. If you liked Iron Man you'll probably like Spider Man. As with many AI optimisations, on average this is probably okay. You'll very rarely get a movie recommended to you that you loath. The safety of that comes from not deviating too far from the comfort zone. Doing well on average is a good business model it seems. It also feels sad and boring.
What these models lack is a sense of discovery. Sure, you can safely move through the marvel catalogue and keep people watching but eventually it gets stale. This is probably best seen on YouTube. Log into a fresh account and you'll be shown dozens of videos on different topics. Watch one and see how quickly your recommendations become more of the same. One or two videos on woodworking and you're in the 'likes woodwork' bucket. Without going out of your way to explore you'll be in that bucket forever. The challenge for these companies is to find a way to let users explore more varied recommendations without killing profits.
Curation seems to be a viable alternative. Rather than optimise for an individual's taste, a curated collection aims to excel in just one genre or category. While curation seems similar to recommendation on the surface, it has two properties that make it interesting. First, curation tends to signal itself quite well. If you're looking at the list of the ten best local Mexican restaurants you've opted into that category. Second, by opting into a category you give the curator permission to present a wider spectrum of recommendations. If the top five teen fantasy books were all just Harry Potter it would be a pretty boring list. Curation gives us the opportunity to explore different aspects of something we enjoy. A good curator will choose items that highlight different key features of a genre. Some Mexican restaurants might do great fish tacos while another is known for the heat of its signature salsa. Not every item in a curated collection is going to be a hit. Often times it's the differences between them that makes these collections valuable. Through experiencing the full spectrum we learn more about what we enjoy. Those bad experiences become valuable lessons.
Somehow we need to find a way to curate on mass whilst not losing the value of discovery. Personalised curation quickly falls back into mimicry and trying to curate for everyone leads to more regression to the mean. Beyond that, hyper personalised curation removes the potential social imprinting that comes from curation. Oprah's book club was no doubt popular for reasons beyond recommending good books. Reading those books and talking about them to others created a tribe. I'm yet to hear anyone start a social club to discuss their Amazon recommended purchases. Whatever the solution there will surely be failed attempts along the way. But even in this, sometimes you need to kiss a few frogs.