Thread · 12 tweets · 1 Jun 2021

Governance in digital advertising? I can hear you mock me from where I sit. Yet I think it's a worth a shot to solve a relatively constrained problem (details in 🧵) and if that works maybe build a healthy ecosystem for advertising on it. My first draft: darobin.github.io/garuda/

One active area of work today is how to enable effective advertising — which is a source of revenue for which there is no clear alternative — without carelessly broadcasting personal data. One important component to achieve that is the ability to anonymise ad requests.

With anonymised requests: ✅ Users get the privacy they want (not some "consent" nonsense). ✅ Publishers protect their audience from being stolen by third parties. ✅ Adtech can operate effectively by reusing a lot of existing infrastructure.

The problem is: who will you trust to run a Web-scale anonymisation service?

Garuda establishes a governance model for a trusted server, operated Web-wide, that can provide several useful services to the advertising ecosystem, chief amongst which is request anonymisation. There are more details in the draft: darobin.github.io/garuda/

Ideally, we could solve this through purely technical measures as those offer stronger guarantees. Basically: technical measures > multistakeholder governance > might makes right But computer scientists haven't solved everything yet, so we can't always rely on TMs.

Operating an anonymising proxy gives whichever party operates it some very strong incentives to defect, betray the user, and steal the publisher's audience. That's why this is built atop multistakeholder governance: you can only get defection if all stakeholders defect.

This approach differs from that which has been preferred to date in digital advertising. What's commonly done there is to somehow organise a pinky-promise bank or mesh with close to no enforcement, and claim that as governance. We've seen how well it works.

Is this an ideal solution? No — institutions are hard, and the spec is only an outline. But we are going to have to solve Web-scale governance, and this is a good place to start.

Does this address all potential points of defection? No — the user agent still remains. For that we need matching policy-level changes, and I have a proposal. papers.ssrn.com/sol3/papers.cf…

Is this issues-free? No — there are clear problems in representing users and the buy-side. If you have expertise in multistakeholder governance, platform coops, etc. and you can see a way forward, please reach out! I've spoken to many but we're just getting started.

At the end of the day, the Web is about trust. And users can trust the Web *because* they don't have to trust each individual website that they visit. The advertising status quo is predicated on the violation of that trust. Governance is a strong way to rebuild that.