The sample audience might work for CPC campaigns on large sites but difficult to see how it would work reliably for the millions of smaller sites out there (or work at all for CPA campaigns where conversion % can vary wildly depending on the action being paid for and the geography of the end user market being targetted).

I'd personally have zero problem with a strongly audited third party tracking me during transition from a publisher site to arrival (or checkout in the case of CPA) on an advertiser's site.

What I do object to is that same tracking info being used across every publisher and advertiser on the ad network (and tracking info persisting for months).

If there was an ad network that enforced a strict set of simple, inert advertising formats and ensured tracking was ephemeral and unique between each publisher/advertiser pair (and not capable of being used to parallel construct my browsing across the whole network), I'd happily accept the compromise. Doubly so if I knew the browsers held a nuclear option of "de-listing" the network in the event of bad behaviour.

Well, for CPA the only thing that matters you already know, then it's up to you to decide who you want to run those ads and if they agree with you and find you trustworthy. That's a 'solved problem' in the sense that for a long time such situations were dealt with using outside auditors. It costs a bit but not so much that it would make CPA impossible, merely less convenient.

Interesting part I got from this discourse: there is no 'one size fits all' when it comes to what is acceptable and what is not. That's yet another problem to be solved.