as someone who is working in ED in Australia, I can say that Lesswrong has been consistently predict the behaviors of my hospital management by 2-3 days.
for example: I’ve read a lesswrong post arguing that covid-19 is serious and people should stock up while my manager was still telling us to just calm people down if asked. The stance was reversed 2-3 days later.
Slatestarcodex recommended the wide spread use of masks a whole week before my local health district recommened the same thing.
I am not certain that being slow is not part of the system design. but Lesswrong has allow me to predict future hospital policies quite consistently.
while I agree with most of what you said and in an ideal world ad should work in a win-win manner as you described, I have cut out as many ads from my life as possible since they are significantly net harmful in my experience.
the problem that I found, and you don’t seem to address, is that ads are not just a simple showing of “I have the stuff you may want”. It is usually an attempt of manipulation using primarily superstimulus or social engineer to maximize profit for the advertisers. e.g. for a car ad they show happy people living exciting lives which have no relation to the car but make you associate the buying of the car with non-existence social fulfillment.
It would be ok if advertisers’ incentives are aligned with ours. But usually, they are not perfectly aligned if not horribly misaligned. And I assume that companies that use “honest” ads would fail to compete against “superstimulus” ads. So the majority of ads would be the equivalence of attempted mind control which rational agent should avoid even at the price of not knowing that there are things you may want to buy.