On Tuesday, 25 June 2013 00:31:11 UTC+1, Sid Stamm wrote: > I don't think we all agree it's the best route, but it's worth trying > out. At least as initially conceived (in a way that blocks all > unvisited sites) it has user-expectation and accuracy problems. Brendan > has written up some of his concerns. > https://brendaneich.com/2013/05/c-is-for-cookie/
While Brenden's posts make some good progress, they're still trying to solve the implementation problem of "Can we block a high percentage of tracking cookies while allowing a high percentage of non-tracking cookies". I'd like to see more discussion about whether the principle of blocking tracking cookies would actually reduce tracking once the trackers had been given a chance to adapt. > > In my opinion it is impossible to prevent a user being tracked from > > one website to another. > > I wouldn't write it off as impossible, but yeah, it's incredibly hard > and would probably greatly degrade the web experience. I think there are three fundamental requirements for a tracking network to operate: 1) You must be able to pass data between parties (e.g. myshop.com and my-analytics.com). This can be done behind the web server, so is impossible to block in the browser. 2) A single site must be able to identify a user as they move from one page to another. This is essential for any stateful website. 3) There must be some mechanism for identifing a user on one website as being the same user on another website. At the moment this is most commonly third party cookies, but there's a huge range of alternatives to use from localStorage through to real name. As far as I can see preventing (1) is absolutely impossible and (2) would break the web beyond recognition. There might be some scope for (3), but even if you managed to remove all the technological means for two websites to identify an individual, as soon as a user starts interacting with a site they're likely to reveal details like their name and email address. > Can we make cookies more useful *and* more transparent? Can we drive > the fair players away from cookies onto some tech that's more useful for > their purposes and also in better control of the users? Maybe we can, but I don't think these proposals do that. They potentially inconvenience all web developers and would likely cause the services we are worried about to use less visible methods. > I don't want to assume users hate tracking, but the opt-in rate to DNT > is staggering -- higher than any off-by-default feature in Firefox. > This means quite a few people want tracking control of some sort. Agreed, but to me that's an argument for tying any setting to Do Not Track, rather than enabling by default. This could be dumbed-down with some form of low/medium/high privacy slider - expose the intent, not the implementation (except to advanced users). > Not sure where you're shopping, but my frequent shopper card gives me > more like 10%, the card doesn't automatically present itself and I know > when the card is used and that it is used by only the grocery store (not > the library or parking meters). So it's already far less subtle than > cookies. This is heading off-topic, but very quickly 1% is standard in the UK and the same card can be accepted by multiple different businesses (I think the widest acceptance is http://www.nectar.com/collect.points). I don't know if they share data, but if not the technology wouldn't be hard. It is definitely more obvious than cookies though. > If you track my purchases via swiped credit card without > telling me it's for more than payments -- that's more like cookies. Which brings me back to (3) above. Ian Thomas (thelem) _______________________________________________ dev-privacy mailing list [email protected] https://lists.mozilla.org/listinfo/dev-privacy
