William Eliot

Hi! I'm a developer and writer exploring the creative side of AI.

Survival doesn't mean quality, right?

08 Mar 2024

People say ‘the cream rises to the top’: supposedly the best people, the best ideas, the best things rise to the surface so that everyone can appreciate them. But we all know that there are some incredible songs in our playlists that haven’t received the attention they deserve, and likely never will. It’s not because there’s something mysterious holding them back from wider success, more a symptom of our personal taste being different to popular taste. That band you love shockingly hasn’t hit the big time yet, but the shock is there only because you’re overrating them in contrast to everyone else. Or maybe they just need better marketing.

So, an empirical measurement of aesthetic quality? I really hope it’s not popularity over time. If the equation for whether something is ‘good’ is just number of interactions divided by how long it has existed, then that means Desperate Housewives is better than Fargo, and Despacito is a good song. I’m sure there are way more parameters in the equation, such as marketing or novelty, but those two parts (popularity and time) would still be the core factors. The implications are quite concerning — come on, Harry Potter can’t seriously be one of the best books ever — but I’ve yet to think up a different measurement.