Had a really small question which felt too localized to ask as a real question, but in the following construction "In principle the server and not the user is in control of the private key, but a user should be able to generate his own private key and use this private key to generate a timestamped signed private key for the server." is there a name for this construction? or for simply generating another secret key from a private key?
was hoping somebody might be able to enlighten me here xD
@DavidMulder way back when there were suggestions to have the ability to follow users and the like. I personally see nothing wrong with this at all since you already can by subscribing to a user's activity - users have a "user feed"
It just tickles me that back then they were adament SE was not a social network and the new profiles remind me a bit of github
I say again, though, I do like them
Really
I just have an odd sense of humour - I am a rhino :)
No, not like this:
Hear me out!
As Stack Overflow grows, it becomes increasingly ridden with extremely newbie and mediocre questions. It's just natural. The team are fighting hard to keep the crap out, but there are many very basic questions that are okay on their own. They don't deserve to b...
I think it would be nice to have a "friends list" on Stack Overflow. This would allow us to be able to follow the questions and answers that are posted by users that we want to watch/monitor.
I suggest it be implemented in the same fashion as DotNetKicks.com; where you can add someone to your "f...
I think the concern was that friends list tends to voting ring quite quickly. Answers should stand or not on their own merit separated from who wrote them
Oh well, what I meant was that one of the founders wrote a post where he described how they consciously decided not to build a community. Because communities work when you're a small tight group, but when they become big they always fall apart.
I'm not sure it's ever explicitly made it to a blog post, but the sentiment has been expressed in a few of their podcasts and other places. I can't remember where it came from
Because StackOverflow (the biggest community) still has all users act quite politely to each other xD And the more social features you would add the more that would probably dissapear (I don't know of a single big social community that works)
yeah, I read it in a blog post, but I think it actually might not have been a blog post on blog.stackoverflow.com (where I was searching just now~)
@DavidMulder Now, for this - are you wanting a private key of the symmetric kind to derive another private key from? I don't know of any construction exactly matching this format (probably worth talking to @CodesInChaos and @ThomasPornin)?
Certainly one way to take user private information that the user knows and generating other keys is key stretching - but this doesn't sign or timestamp keys
And usually the idea of this is to derive a symmetric key that is not weakened by the fact you tried to use an 8-letter ascii string for the key thereby weakening it from 2^128 or whatever to 2^not very good
The process almost sounds like a certificate signing request (pkcs#10)
you generate your keypair, then you ask a certificate authority to sign it for you
the resulting "certificate" is then trusted because it is signed by the certificate authority
you can have as many levels to this as you want - called intermediate CAs
thus there's no reason you can't have World Dominating CA issue a cert to Your IT Team, who then issue certs to servers
there might be a cryptographic construction for what you're looking for, but... I'm not sure there is. On the other hand, I caveat that with Thomas and Codes are the people to answer that authoritatively, not me
and although my knowledge of security in general is pretty strong (as in, as far as programming stuff goes and all the different 'normal' ways applications can get attacked) my knowledge of cryptography isn't. The problem I am facing is that I have a distributed network of servers, where every server has users and each user is given a private key and the public key is distributed (to allow signing of messages that get sent).
The problem is when a user migrates from one server to another that he should be able to prove which one he is. So that's where I got the idea that a user could optionally (only if he requires the extra security and flexibility) generate a private key privately and generate the 'shared with the server' timestamped private key in a way that would still allow it to be checked with the old public key.
I am pretty sure this is theoretically possible (as long as you also know the old 'shared with the server' private key), but I have been having a real hard time putting it in correct terminology in my specification of the platform xD
Well, I am fine with moving the private key, but in the simple case where I move the private key the old server owner could still impersonate the user on the network
For the sake of simplicity that will be possible for 'normal' users, but I want to create a more robust system for users that have more strict requirements (similarly how users can optionally enable client based encryption with off-the-wire shared passwords whilst normal users will just rely on the basic security of the network)
That's my problem, I really should do more reading in the area of cryptography xD I think I have got a lot of the practical stuff down quite well, but my knowledge of the theoretical parts is far too weak :(
I think you should ask this on crypto - it's a definite protocol problem and definitely within scope. If you can get all this detail down specifying exactly what conditions you want it should get an answer
Might be worth searching there too, see what there is
This is not as simple as straight up PKI by any means
And if you just want to stick to chat crypto has a chat room :)
@Rhino Yep it is~ :P And true that, that's why I have now gotten into the habit of writing out questions I have in notepad and then waiting awhile before actually posting them xD Either in the process of writing them or when working on other stuff you find answers so often :P
I have some ideas, but I'm really hesitant to suggest them. I'm not an expert protocol designer and I'd love to see what the actual experts have to say.
The thing with crypto is you can so easily be utterly and totally wrong... anyone can design a system s/he believes is secure and all that
This is my weekend destroyed. The rest of it will be spent reading up on key agreement protocols :( :)
Haha, I am half afraid that it will end up the way some mathematicians feel when they end up on mathoverflow :P And yeah I know@last thing you said is so very true. I am now working on a global spec where I know that all the requirements I am writing down are theoretically possible, but if I get down to actually implementing the cryptographical parts of the system I am definitely not going to do that without a lot of expert input
Just need to get it all on paper now (as in, the coming two months) so that I can convince the right people xD been dreaming about this platform for the last 3 years and wrote 15 pages with just general outlines in the last 2 days xD
@Rhino Hahahahaha, like mathoverflow is the only site on all of SE I don't even bother opening, I just don't get it xD like not at all xD
give me physics, biology, puzzles or philosophy and I feel right at home, but mathoverflow and people studying maths I just have so much respect :D
@DavidMulder I usually go through three phases 1) Oh I think I vaguely understand this one post in 200! Yes... 2) Ok that was more involved than I thought... ok well... 3) reads answers... oh crikey I have so much to learn
There is a formal definition for the notion of a formal proof.
Question 1. Is there any formal definition for the notion of a diagonal formal proof?
Consider the following theorems both proved by diagonal proofs.
Theorem 1. (Cantor) $2^{\aleph_{0}}>\aleph_{0}$
Theorem 2. (Godel) $ZFC\nvdash C...
@Rhino I am curious though, you as someone who knows A LOT more about maths than I do, do most of those questions asked on MO look like they are by people doing actual real-world work (programmers, econometrists (or whatever way you say that in English)), or are those mostly university hired mathematicians or similar theoretical people
and then I read the answers and feel very small and insignificant
@DavidMulder It's research mathematics. Originally, SE 1.0 was a standalone site started by research mathematicians when you could get an SE site for whatever you wanted
SE 2.0 was where they made area51 and central control and community voting
MO is pretty much the only survivor
so yes, these are postdocs and professors discussing actual questions for their research
not obviously their whole research, just bits and pieces
@Rhino Yeah, I get the stuff on maths.SE :P Can't answer it, but I do get it xD That's why I specifically mentioned MO (I have another friend who just started studying maths and he told me he didn't get MO at all, but he couldn't answer my previous question as he is only in his second bachelor year now xD)
Correction, after looking again, I get about half the questions on maths.SE xD
@RoryAlsop You could have changed it to something like "Hello I just wanted to say I love you all". It doesn't read as sincere with the huge "heeeyeyyy" in front of it
@Ohnana could do. Depends on who you ask. Technically - and I say technically given the content of this room - the standard of behaviour here should be "professional"
StackOverflow recently migrated to a new profile dashboard. Are security.SE admins planning to upgrade as well?
There is a formal definition for the notion of a formal proof.
Question 1. Is there any formal definition for the notion of a diagonal formal proof?
Consider the following theorems both proved by diagonal proofs.
Theorem 1. (Cantor) $2^{\aleph_{0}}>\aleph_{0}$
Theorem 2. (Godel) $ZFC\nvdash C...
