Sven Slootweg on Nostr: This is not a "harmful generalization". This is literally *how the advertising ...

This is not a "harmful generalization". This is literally *how the advertising industry works*. It *is* the "actual problem". Why do you think that practically every single advertising platform has some kind of bidding mechanism nowadays?

It doesn't matter whether you can think of scenarios where it "isn't that bad", regardless of how "realistic" they are. You measure the impact of a system on society by measuring its *worst-case* scenarios, not its *best-case* scenarios. And that is what I am trying to illustrate here.

The model we have for advertising is optimized *for* that worst-case scenario, in that it always converges to it at scale. This is a choice as a society. We can choose not to accept that, and instead pick models that behave respectfully towards society *even under worst-case conditions*.

But to do that, we first need to acknowledge that this model is bad, and that it's not actually optimized for discovery; it's optimized for the *suppression of* discovery, and the "discovery" component is just used as a moral fig leaf.

Once this is understood and acknowledged, we can select models that *actually are* optimized for discovery, without this worst-case failure mode.