Randomness is an underdog.

Serious question is how often would you tolerate if those randomly displayed posts are absolutely out of your interests? Would you click or skip? Plenty of fish (Canadian-based dating site), programmed by Markus Frind, had a function: during onboarding you could choose types of people you think you prefer (e.g. brunette/blond etc.) and if you haven’t clicked later on them, algo had started to show different results…

Interesting anecdote on Plenty of Fish. It’s definitely interesting how people aren’t really good at telling you what they like; empirical evidence is far better. I believe Paul Graham has an essay on a similar topic where if you ask people if they like an idea you have for a product, they are likely to say yes even if they wouldn’t actually use it. But if you ask them how much they would pay for access, or if they’d pay a certain amount, you’d get a more accurate response.

FWIW, I wasn’t suggesting pure randomness though, it’s more like probabilistic randomness. Rather than a binary threshold a post must pass to make the homepage that divides the community into curators and consumers, this would show you posts with a degree of randomness with a probability proportional to the likes it’s garnered.

Btw, I’m not sure what you meant by randomness is an underdog? Are you implying it’s a nice goal but it rarely works out in practice, perhaps because people actually do fall into natural curator / consumer buckets?