The key premise of algorithms is efficiency. Algorithms optimise data for decision-making that we previously did not have before. But humans are not efficient. We are messy and inherently flawed, which makes us human. We value serendipity, community and experiences, which may not be efficient but is impactful.
If you go too far off the beaten path of algorithmic consistency, you risk never being heard. You may never get enough cultural momentum to exist in any meaningful way. That is a real problem. But if you're too good at algorithmic optimization, you risk building a large audience without saying anything particularly original or important. Not to... See more
The data informs and influences us, but we choose based on the totality of our experience
learning to use data properly
McNamara Fallacy: A belief that rational decisions can be made with quantitative measures alone, when in fact the things you can’t measure are often the most consequential. Named after Defense Secretary McNamara, who tried to quantify every aspect of the Vietnam War.
I believe the “ tyranny of numbers ” has atrophied people’s abilities to listen to their body. Numbers can be helpful, but often cause me to lose the plot. I’ve found that I feel better when I let my body tell me what I really want, instead of the non-actionable anxiety associated with metrics.
When everything is measured in terms of engagement, content will be optimized for addictiveness. Not entertainment or artistic merit. Not intellectual substance or creativity. Pure, plain addictiveness.
I’ve struggled with this balance myself. When the audience responds to something I’ve written, it can be tempting to chase the revenue and viewership by repeating the format. Repeatability and audience fervor can build a business, but kill the artist.
The Unaccountability Machine by Dan Davies is a fantastic book, which argues that people create these models because if you can reduce decision-making to an algorithm, or a formula, or a process, or a procedure, you avoid the risk of blame.