The spam analogy for targeted ads is a really bad one, I think. Spam became popular for senders because it was extremely cheap to reach millions and millions of

The spam analogy for targeted ads is a really bad one, I think. Spam became popular for senders because it was extremely cheap to reach millions and millions of people. Over time costs did grow a bit, because of the effectiveness of filtering, but at the same time if it's unlikely for your spam to be ready, how much would you be willing to pay for it? Not that much, hopefully. Targeting does not necessarily reduce the cost of ads - for very valuable users the cost of an extremely targeted ad can be quite high. The question is then what happens to the total cost of the campaign. The article seems to imply that the cost of the campaign is low for low-quality sellers, and this is why targeted ads become a bad signal for buyers. But online ads are very often performance tracked, and the entire reason that some demographics are so expensive is that the ads to them perform well. If an ad is measurably giving a high-return on investment, then the advertiser should be pouring as much as they can into the campaign, which implies that campaign budgets for some targets go up, not down, and the quality signal to buyers should be higher, now lower. So it seems like as targetting differentiates by consumer, it also differentiates by seller. The ad networks also allow choosing which publications to target, and higher quality publishers do cost more. So given some self-awareness about your own value for a given target and the quality of the publication you're reading, you could (given the signaling theory of ads is correct) get a better signal in many cases.