ekr____

16 ms·

241.
by ekr____ 4y ago
As far as I can tell, mostly not. The data is all signed, so if the user mirrored a copy then they would be able to upload it somewhere straightforwardly, but it's not a built-in feature of the system as with (say) a DHT.
242.
by ekr____ 4y ago
To be honest, I meant to mention MASQUE as an aside, and it's in my notes somewhere to put in but it seems to have gotten lost. Anyway, I think it's great that we're seeing some new tech here, but it's of course been pos
243.
by ekr____ 4y ago
Good question. IP addresses are used for routing, which requires them to be structured. The basic idea here is that your ISP advertises a route that consists of a "prefix", i.e., a block of addresses it controls. So, for instance
244.
by ekr____ 4y ago
Assuming you're talking about RFC 8981, then I don't think this actually helps as much as you would like, because typically you would have an assigned /64 and then just randomize the low order bits, but the /64 would be
245.
by ekr____ 4y ago
It's not me.
246.
by ekr____ 4y ago
Post author here. Happy to answer questions, take criticisms, etc.
247.
by ekr____ 4y ago
OP here. Happy to answer questions, field disagreements, etc.
248.
by ekr____ 4y ago
Post author here. Happy to answer any questions!
249.
by ekr____ 4y ago
A variety of ways. However the counting itself is usually done in the open which means that people can observe that (1) the counting is done correctly and (2) is then transmitted and reported correctly.
250.
by ekr____ 4y ago
Do you mean on HN or in real elections? In real elections, the situation is actually quite complicated depending on where you live. Specifically: - Many people vote on paper with hand counting, which is of course not digital. - Many people
251.
by ekr____ 5y ago
I don't know for sure how the Fx stack presently behaves, but I would expect that eventually it would do DoH/3 if the server provided the appropriate indications (e.g., Alt-Svc).
252.
by ekr____ 5y ago
Agreed. Basically all you need is: 1. Something to delimit the boundaries of the message. 2. Some way to determine the nonce. OTOH, the TLS 1.3 record layer is barely more than authenticated encryption,
253.
by ekr____ 5y ago
I know that tptacek knows this, but for those who don't, we took data compression out of TLS 1.3!
254.
by ekr____ 5y ago
The primary reason you need records is that AEAD algorithms run on blocks so you need to know where the end of one block is and the start of the next. And because you are running over TCP, you don't get that framing information from TL
255.
Understanding the Web Security Model, Part II: Web Applications(educatedguesswork.org)
2 points by ekr____ 5y ago|0 comments
256.
by ekr____ 5y ago
It took me a little time, but here's a hopefully readable description of how it works: https://educatedguesswork.org/posts/ipa-overview/
257.
by ekr____ 5y ago
You're right that this assumes an account. The way that cross device/browser identification works in IPA is that if you are logged into example.com (though realistically, often Facebook) then it can set a "match key" whi
258.
by ekr____ 5y ago
The naming around the various features is confusing. Briefly: - Tracking Protection basically blocks connections to known trackers entirely so the resources don't even load. - Enhanced Tracking Protection prevents various kinds of cros
259.
by ekr____ 5y ago
As I noted above, this is just a document at the moment, so it's not in Firefox at all. With that I said, ublock origin blocks ads before they are loaded, so I would not expect an API of this type to affect it.
260.
by ekr____ 5y ago
What "partner" in this case means is "Mozilla engineers collaborated with Meta engineers on developing this proposal". We also routinely partner with Google engineers on technical proposals, see, for instance: QUIC Versi
261.
by ekr____ 5y ago
There was a discussion in PATCG Wednesday about precisely this topic. Briefly, it should be possible to "disable" this feature by instead encrypting a random match key, which would look the same as having it enabled but merely no
262.
by ekr____ 5y ago
They have similar objectives but achieve them differently. Roughly, Private Click Measurement uses very low entropy identifiers to achieve rough of clicks and conversions. IPA uses multiparty computation instead. Probably the most salient d
263.
by ekr____ 5y ago
Two clarifying points: 1. This is a technical proposal. It has not been incorporated into Firefox; I don't think there are even patches for it at this point. 2. Safari actually already implements a different technology designed to achi
264.
by ekr____ 5y ago
No. Topics is oriented at targeting ads to people with specific interests. This technology is designed to measure "conversion", e.g., when people click on an ad and then buy the corresponding product in a privacy-preserving way.
265.
by ekr____ 5y ago
I think a more likely way to think about it is that this allows us to sandbox things that would otherwise would not be sandboxable. For a variety of reasons, it's probably not practical to remove the existing process sandboxes.
266.
by ekr____ 5y ago
Original author (though not OP). Much of this should be old news to people here, but happy to answer questions if people have them.