Reading this, it would be interesting if we had some standard Privacy Tests for browsers. Much like perf & security, you take stock browsers and run them over a set of pages not known in advance, in a modified environment. Look at what happens on the network, disk activity, etc.
DHH
 @dhh
10 Dec 2019
“Since Apple introduced what it calls its Intelligent Tracking Prevention feature in Sep ‘17.. advertisers have largely lost the ability to target people on Safari based on their browsing habits with cookies, the most commonly used technology for tracking” daringfireball.net/linked/20…
1
2
Publish the results on a regular basis. See how well all browsers do without turning knobs at protecting users from tracking, privacy leaks, fingerprinting, ad exposure, and the like. The pages could exercise common vectors, but vendors couldn't optimize for them.
1
A complicating factor here, in terms of the benefit for users, is that how much sites will work around these things varies depending on the market share of the browser. They'll do things for a browser with 30% share but might let the users of a 3% share browser go untracked. 1/N
1
For example, a browser with larger market share that makes cross-site tracking ineffective will make a lot more sites move to fingerprinting than one with smaller market share. 2/N
1
This means that larger-share engines need more thorough modelling of privacy like a security attack surface for privacy (or similar) mitigations to be equally effective (share of sites, time decay of effectiveness). 3/4
1
1
Replying to @davidbaron @humphd
Whereas engines/browers with smaller shares can *sometimes* get real wins for their users, at least for a decent length of time, by blocking just the current technique rather than the entire underlying problem. 4/4

Dec 10, 2019 · 4:21 PM UTC

1
2
Replying to @davidbaron
So maybe expanding this to look at data from different angles is important, too. We have users, sites, browser vendors, etc. each with something different to understand. Re: 3% browser and market, maybe also being able to see data over time helps with this vs. single point
1
I still think there's value in examining this from the standpoint of a user, since user's have few things they can do (choice of platform, perhaps, browser). Having *any* understanding into the trade-offs of that choice would help. We know a lot about speed/compat vs. privacy.