Thread by @Myles_Younger

1/ I wanted to note several thoughts on TURTLEDOVE, a proposal to have ad auction/targeting logic execute in-browser. I think it's an awesome thought experiment and contains elements that may survive (read to very end), but overall I think it's fanciful. github.com/michaelkleber/…

2/ This thread contains really specific quibbles, but they raise key questions if we (digital advertising) think that we can basically hack our way back into The Good Old Days via —in this case— browser-native functionality. TURTLEDOVE feels like a hammer in search of a nail.

3/ But this isn't really about TURTLEDOVE specifically. This isn't meant to ridicule the concept. My point here is that trying to preserve The Good Old Days might be a mirage. Precisely because it is so detailed, TURTLEDOVE is great to examine.

4/ I'll go through the Github doc bit by bit.

First thing is in the introduction. My question: isn't this ALREADY what a browser does? Browser history, caching, etc. What TURTLEDOVE asks is for users to make a major concession for no one's convenience but that of advertising's.

5/ Then there's this bit. From the user's perspective, they're still being stalked on an ADVERTISER'S TERMS. The user—an average citizen—isn't going to care that the targeting was locally-executed. To a consumer, it's still going to feel like being chased.

6/ A little further down there's some suggestion of contextual targeting. This isn't a net-new capability. In fact, contextual is ALREADY poised to be one of the strongest surviving tactics in the wake of the Cookiepocalypse. TURTLEDOVE isn't needed for this.

7/ Getting to the end of the Introduction. Not provable either way, but this is perhaps one of ad tech's greatest fallacies: that people somehow "prefer" certain ads (ads delivered by ad tech, naturally). I don't think that's axiomatic. Not by a long stretch.

8/ Next sentence. Another questionable axiom. It can be argued that ad tech and specifically its "race to the bottom" economics have NOT been good for publishers (and their readers) at all. Could also be argued the other way. My point is this can't be taken for granted.

9/ This is also how marketing worked before cookies and ad tech. Marketers don't need TURTLEDOVE to target interest groups. This is another case seemingly of a hammer in search of a nail.

10/ So we're just trying to find a way to resurrect site retargeting without 3P cookies?

11/ Retargeting cart abandoners. Oldie but a goodie I guess. I mean, there's a whole side rant one could make about the role that flawed last-click attribution played in retargeting's rise to prominence.

12/ This is mostly true. But it only MATTERS if browser users are sitting around saying: "You know what? I really liked those shopping cart retargeting ads; I just didn't like the invasion of my privacy." (I think the saying "Said no one, ever" may apply here)

13/ OK, getting into the tech parts. Wait...wouldn't first-ad-network.com get the IP address of the browser AND a domain the browser visited? There are other ways to do this (e.g., thru proxy endpoint), but getting iffy on PII leakage. Note: author DOES address this later on.

14/ Now THIS is why I have (half) jokingly referred to on-device approaches as "the return of adware." We're now INJECTING Javascript into the user's browser. If ad tech has demonstrated anything, it is that any possible loophole will be exploited, and this seems ripe for misuse.

15/ You didn't ask for it, but here's more on "the return of adware." The funny part is, nearly two years later my "Weird Outcome no. 3" is actually seeming pretty plausible! linkedin.com/pulse/three-we…

16/ I have trouble understanding why publishers would want to go along with this. They're finally regaining some power and now they're going to go right back into a situation where every ad impression is a coin flip they have no control over?

17/ This is seeming pretty complicated and contrived. If you want to preserve the most annoying parts of behavioral advertising while protecting people's privacy, I guess you have to twist yourself into a pretty knotty Logic Pretzel.

18/ I just feel like TURTLEDOVE is over-engineering something that can (will) be solved with contextual buys, publisher first-party data, etc. It just...feels over-engineered for what the ecosystem (and users) would be getting out of it.

19/ This I like. The notion that the browser essentially maintains its own firewall—total departure from ad tech's foundation of unlimited 3P HTTP calls. Good for ad tech? No. Probably a next salvo in the Browser Privacy Wars? Yes.

20/ This I also like. The notion of caching or batching ads to keep the user at arms length from advertisers and networks. This may be a solve...but I'm not 100% sure it would happen within the browser. Who wants to use a browser that's basically adware?

21/ Seriously, how much disk space & CPU cycles would all of this take? I'm not kidding. Think about taking the collected machinery of ad tech & cramming it onto your own machine. This means YOU (the user) are now using up YOUR disk space & CPU cycles to help...ad tech? Really?

22/ OK so here the author covers the whole IP address leakage thing. Yes, that part is def. solvable. But with added complexity! Once again, trying to preserve the Good Old Days is requiring layer after layer of (over) engineering with no clear benefit to the user.

23/ I will say, though, and I have noted this before, that I think some sort of native VPN or IP address obfuscation is a likely feature we'll see from the likes of Safari, etc.

23/ I think I've made my point. I think this concept may have legs, just NOT IN WEB BROWSERS. I think the user expectation of how a web browser ought to work diverges too far from what TURTLEDOVE aims to do.

24/ Ah, but you start to think of set-top boxes, smart TVs, smart speakers, and the panoply of other internet-connected devices. Maybe these would be more suitable homes for local execution of ad targeting. Mostly because it just wouldn't be as obvious/onerous to the user.