5 ms·
I understand your point, but the problem is that anybody can make their DNS response say whatever they want. And you can't even rely on crypto signatures (e.g.
by p4bl0 7y ago
I understand your point, but the problem is that anybody can make their DNS response say whatever they want. And you can't even rely on crypto signatures (e.g. DNSSEC), as the people controlling some DNS records are not necessarily the same as their users.
For example, GitHub controls DNS for *.github.io so if they wanted to track people across all GitHub Pages, they could.
Also, the Public Suffix list isn't something that needs to be polled every time. It is reasonable for browsers to cache it for several days even.
- quotemstr 7y agoI don't think we'd in practice see people lying one way or the other about a particular name level being "public". If GitHub wants to track people across all *.github.io pages, it can, trivially. It controls the infrastructure, after all.
- jefftk 7y ago> If GitHub wants to track people across all .github.io pages, it can, trivially. It controls the infrastructure, after all.* If WebKit fully implements what they're describing here then GitHub should not be able to use their control of *.github.io to track people across those pages.
- p4bl0 7y agoquotemstr was talking about the infrastructure that they control. It's true that they can track visitors server-side if they want to.
- jefftk 7y agoWith what WebKit describes here GitHub won't be able to tell the difference between: A: a browser visits both foo.github.io and bar.github.io B: one browser visits foo.github.io, another visits bar.github.io Both should look identical to github.io, client side and server side. (This is not the case today, but my reading of the policy is that they consider all the ways in which it is not the case to need fixing.)
- p4bl0 7y agoWhat about the IP address and the other information that leaks at network levels that are lower in the stack than what the browser controls?
- jefftk 7y agoMany people are behind NATs and so share IPs. This can be millions of people; see Wikipedia's documentation on IP banning: https://en.wikipedia.org/wiki/Wikipedia:Blocking_IP_addresses https://en.wikipedia.org/wiki/Wikipedia:Blocking_IP_addresse... What are you thinking other than IPs? Everything else should be under the browser's control.
- p4bl0 7y agoI don't know. I just suppose that it's not that simple. It is never a good idea to assume something is safe for good. The TTL of the packet could also be used maybe? Also, there are also many people that are not behind a NAT. I've always had a public IP address at home, for example. Plus NAT may become less used because insert rant about NAT and IPv6. Anyway, it's good that browsers make all they can to ensure privacy :).
- jefftk 7y ago> For example, GitHub controls DNS for .github.io so if they wanted to track people across all GitHub Pages, they could.* The public suffix list historically has been opt-in. If you look at the (human readable) https://publicsuffix.org/list/public_suffix_list.dat https://publicsuffix.org/list/public_suffix_list.dat you'll see: // GitHub, Inc. // Submitted by Patrick Toomey <security@github.com> github.io githubusercontent.com Major companies have historically wanted to list their user-content domains on the PSL because it helps protect users cookies from other users. It's possible that this announcement changes things, but I think it mostly doesn't. The main problem with a DNS-based system is just that you would need to get everyone to switch to it, and figure out what to do in the mean time. If ".io" or someone doesn't set the DNS flag then that's a major security issue for everyone under it.