Ad networks do three things to boost efficiency: 1. Attribution 1. Tailoring 1. Detecting fraud An ad network without any of these will be less efficient th

Ad networks do three things to boost efficiency: 1. Attribution 1. Tailoring 1. Detecting fraud An ad network without any of these will be less efficient than its competitors. It will attract fewer ad buyers. Currently, all three of these things require tracking. Ad networks can use attribution to detect some fraud on popular sites & apps. This doesn't work for the long-tail of small sites and apps that don't generate many clicks. Ad buyers could buy ads directly from popular sites/apps. This just shifts the burden of detecting fraud from the ad network (which has the expertise to do it) to the ad buyer. This excludes small sites & apps, slowing the rate of innovation. The proposals for privacy-preserving ad tailoring all provide coarse tailoring (example: "interested in bicycles"). They do not allow precise tailoring like "interested in bicycle brand X product model Y". Precise tailoring is very efficient. I welcome all work on privacy-preserving ad-tech for these three areas: attribution, tailoring, and fraud-detection. It doesn't matter who develops a proposal. Let's examine the proposal and evaluate it on its merits. Eventually, we may end up with some combinations of techniques that allow efficient privacy-preserving ads.