Alright, I'll bite:
> it's hard to believe that people would rather have irrelevant ads to relevant ads
I usually prefer irrelevant ads.
It's possible for ads to fulfill their idealized role, where they inform me about an opportunity for a mutually-beneficial purchase. But in practice, I view most advertising as basically adversarial; most ads want every viewer to buy, whether or not they will actually benefit, and so they're incentivized to manipulate me. Fundamentally, everything beyond "this product with these traits exists at this price" tends to be manipulation. Grabbing attention, building brand associations, creating demand, spreading FUD - these are all attempts to extract money from some marginal buyer who won't benefit, but can be tricked into thinking otherwise.
Targeted advertising doesn't change this. It does raise the percentage of mutually beneficial ads, and allows ads to get attention with relevance instead of distraction. I'm grateful for those things. But it also helps advertisers build models with which to do a better job of manipulation. My targeting preferences are basically polar; "random billboard" and "recommendation from a close friend" are both preferable to "some targeting".
(This is not actually the core objection, though, which is that I don't trust the firms behind the targeted advertising. Even when targeted ads are better than untargeted, I expect to gain much less than I lose from the advertiser using/abusing/losing my data.)
> if they want to aggregate large amounts of data about internet browsing habits to get a better idea of the types of customers which want certain things, I see that as only a positive since it helps steer businesses in the service of their customers
Again, this seems like an incredibly optimistic view of what businesses do with consumer data.
If Ford wants to aggregate Car & Driver browsing data to decide which features to put in their new model, I agree that this probably benefits consumers. But if United sees that people who search for similar flights multiple times are more likely to book, and then charges me higher ticket prices when I do so, the targeting has been purely harmful to me.
We don't have to resort to hypotheticals like being denied health insurance based on search data. It's already the case that browsing data is not just used to make aggregate decisions, but combined with tracking in ways that consistently harm consumers.
> That is, people are worried about Google telling Capital One your IP address is in the market for a credit card while hackers are walking out the back door with hundreds of thousands of social security numbers, names, addresses, and credit scores.
My fear is that Google will tell Capital One my IP is in the market for a credit card, they'll talk to Experian to see if I'm a good bet, and then hackers will have that much more information when Experian loses all of my personal data. Which they did. Or that when Capital One gets hacked, they'll have not just my SSN but a history of IP addresses linked to me, which can be geolocated to beat the "which of these addresses have you lived at" challenges which my state foolishly uses during identity verification.
This is my fundamental problem with targeted advertising. It's not about what ads I see, or even about what Google knows. It's that every time personal information spreads, it worsens the risks of all the other times it's spread. Huge amounts of data are already gone, and we can't get them back, and that's a major problem. But it's already standard practice for people selling stolen info to download new breaches and correlate that information to create a more valuable product. Tossing personal info into the seedy world of ad networks makes all the other problems a little bit worse.