Kyle Chayka

I have mixed feelings about this book. The first portion reviews how algorithmic recommendations power social media sites such as Facebook, YouTube, and Spotify. There's nothing especially new, but it is a good summary for readers who don't regularly follow social platform news.

All of these social platforms are for-profit businesses. Their algorithms are designed to keep you clicking and watching, so that the business can profit from your attention. The stream isn't necessarily good content, but it's profitable. Harmful and disturbing content can tick all the boxes and be promoted to users throughout a site.

The relentless focus on popularity skips over niche and unusual work in favor of an algorithmic slush of whatever has the broadest, most generic appeal.

curation, qualitative and quantitative

Next, Chayka contrasts algorithmic feeds with human-recommended collections, using examples from an experienced radio DJ and an art museum curator. This is qualitative curation, instead of data-driven quantitative curation. The author clearly prefers the human-powered version, but I don't buy his argument that it's inherently better.

The algorithmic feeds definitely have problems. But are computers really the only problem with these feeds? What about the profit incentive that drives the algorithms' definitions? Algorithms are capable of identifying unwanted content fairly well when the business is motivated to make it happen. For example, Instagram is well known for its heavy-handed censorship of female nipples, but not male ones.

Human decisions lurk behind every element of a feed. Someone decides to weight angry reactions over happy ones. Someone decides to push certain movies or music not because I might like them, but because there's money to be made.

curation as gatekeeping

When does curation become gatekeeping? Filterworld doesn't focus on that question, but it's definitely what I thought about while reading. Chayka takes time to show how the algorithmic curator picks winners and losers, deciding which content audiences are allowed to see. Commercial success depends on catering to the algorithm.

The author prefers human curation, but human curators can also be gatekeepers. His example of an art museum curator is relevant. Art buyers chase the cachet of owning work by an artist who's also represented in museum collections. Another recent read, Get the Picture by Bianca Bosker, is an inside look at the fine art world that really opened my eyes to the way that artists are dependent on the approval of gatekeepers to survive.

Human curators aren't immune to being influenced by profit, either. The payola scandals of the prime radio era are a fine example, as is the recent exhibit of falsely-attributed Basquiat works at the Orlando Museum of Art.

my personal bottom line

I'm not convinced that human curation versus machine curation is a critical issue. Instead, I want to be free to choose the type of curation that suits my needs, and to be able to opt out of curation at any time.

Transparency in curation methods is relevant too. Paid placements and financial incentives should be clearly disclosed, similar to the way that news journalists routinely note potential conflicts of interest in their reporting.

Social platforms like Facebook, Instagram, and Spotify regularly make it difficult or impossible to avoid their algorithmic filters. What is a "following" list, but my own personal curation? I should have the right to opt out of their filter bubble in favor of my own decisions.

originally published at https://groddle.com/blog/2024/apr/03/… (more)