[00:01:13] *** Joins: wolfcore (~wolfcore@unaffiliated/wolfcore)
[04:49:02] *** Quits: wolfcore (~wolfcore@unaffiliated/wolfcore) (Ping timeout: 244 seconds)
[05:12:51] *** Joins: wolfcore (~wolfcore@unaffiliated/wolfcore)
[06:02:54] *** Joins: Xaradas (~Xaradas@vpn.aslteramo.it)
[06:54:01] *** Quits: Xaradas (~Xaradas@vpn.aslteramo.it) (Quit: Going offline, see ya! (www.adiirc.com))
[21:06:05] *** Quits: baristaTam (~baristata@95.85.34.133) (Quit: WeeChat 0.4.2)
[21:15:58] <cold_sauce> What's the best way to learn about how to implement end to end encryption? I've been googling but haven't really found anything meaningful
[21:16:29] <cold_sauce> I have very sensative data that will be going through my server and I don't even want to know about it at all.
[21:16:51] <cold_sauce> but I want to keep it for another person to be able to decrypt it if they have a key they set up with the original person
[21:27:47] <clinth> well, try to implement as little of it yourself as you can
[21:28:01] <clinth> it depends on what you're using it for
[21:28:03] <clinth> what's the context here?
[21:42:53] <cold_sauce> Bob wants to send Alice a message "hey bby, how r u??" and they've intrusted me to keep this a secret so I really don't want Charlie, Alice's stalker, to see this message. Charlie has a couple 0days and can hack in my server at any time. (I'm assuming) So I want to have end to end encryption so that even the server has no idea what the message is. The biggest problem is that I have no idea how to do key exchange
[21:42:54] <cold_sauce> that is not vulnerable to a mitm attack
[21:43:15] <vishwin> considered PGP yet?
[21:43:17] <cold_sauce> and there's no way they can do it P2P, either. Because Alice is literally never online. Ugh.
[21:43:28] <cold_sauce> so tbh, I know very little about crypto
[21:43:30] <cold_sauce> I'll look into PGP
[21:43:46] <vishwin> I mean, PGP has been around since right before we were born at least
[21:43:58] <cold_sauce> I've heard of it in my FOSS course
[21:43:59] <vishwin> probably the easiest implementation is gnupg
[21:44:08] <m0shbear> isn't anything but in-person as a side channel authentication technically vulnerable to some form of mitm?
[21:44:23] <vishwin> well PGP is public/private keys
[21:44:51] <cold_sauce> ooh, this is different from GPG
[21:45:05] <cold_sauce> that's what I was thinking you said
[21:45:39] <cold_sauce> oh
[21:45:43] <cold_sauce> it uses PGP
[21:45:46] <cold_sauce> well shows how much I know
[21:45:57] <vishwin> yup, gnupg is an implementation of PGP
[21:46:23] <cold_sauce> so should I just integrate my service with keybase
[21:46:28] <vishwin> a lot of people on open source projects' mailing lists use gnupg to sign but not encrypt their emails
[21:46:29] <cold_sauce> as the key server?
[21:46:38] <cold_sauce> yeah, that's why we learned about it
[21:46:40] <vishwin> you can just use one of the big public keyservers
[21:46:41] <cold_sauce> and also to sign packages
[21:46:45] <cold_sauce> ok
[21:46:50] <vishwin> MIT comes to mind
[21:47:11] <vishwin> http://pgp.mit.edu/
[21:47:11] <yossarian-bot> Title: MIT PGP Key Server
[21:47:16] <vishwin> you just submit the public key
[21:47:29] <cold_sauce> if there are possibly going to be over 100,000 users of this service (who knows, tbh), is it reasonable to use that server?
[21:47:57] <vishwin> yeah MIT and a few others are the de facto big public keyservers that gnupg downloads keys from by default
[21:48:00] <m0shbear> planning for 100k from the outset smells of premature optimizaiton
[21:48:05] <cold_sauce> true
[21:48:11] <cold_sauce> I'm just wondering before I commit to something big
[21:48:42] <vishwin> as long as a public key is submitted to one of the big public keyservers, gnupg can download it
[21:52:17] <cold_sauce> awesome
[21:52:18] <cold_sauce> thanks
[21:54:07] <cold_sauce> wait so getting back to my question, would the best way to verify that the key matches the identity is to use keybase?
[21:56:35] <vishwin> each PGP key pair has a fingerprint
[21:56:42] <vishwin> which is already unique
[21:56:57] <vishwin> that fingerprint alone is enough to know who it belongs to
[21:58:06] <cold_sauce> but how does Alice know that that fingerprint is Bob's fingerprint
[21:58:08] <cold_sauce> and not Charlie's?
[21:58:19] <vishwin> they are *unique*
[21:58:47] <cold_sauce> hmm. but what if charlie says "hey I'm Bob, here is my unique fingerprint?" maybe I'm not understanding this correctly
[21:59:41] <vishwin> charlie would need to be able to use bob's private key for that to work
[22:00:09] <vishwin> which would only happen if bob is an idiot
[22:00:20] <cold_sauce> oh I see. yay!
[22:00:32] <cold_sauce> all right. time to learn all this stuff
[22:00:37] <cold_sauce> thanks. you guys were a huge help
[22:17:42] <cold_sauce> what if charlie just gave his public key and his unique fingerprint which was signed by his private key and said "I'm bob"
[22:39:26] <woodruffw> your keypair doesn't prove that you're an actual person, it just proves that you're a pre-stated identity
[22:40:03] <woodruffw> if charlie generates a keypair as bob, then his identity is bob as far as the cryptosystem is concerned
[22:46:18] <cold_sauce> but with keybase, you have to verify it all with twitter, facebook, etc
[22:46:26] <cold_sauce> that solves that issue, right?
[22:46:34] <cold_sauce> you have to verify you are who you are
[22:46:43] <cold_sauce> with the sum of most of your public identities
[23:03:14] <woodruffw> cold_sauce: in the sense that those are all identities as well, but actually verifying that someone is who they say they are online is notoriously hard
[23:03:28] <woodruffw> for example, i could easily create a fake facebook account, fake twitter, fake HN
[23:06:20] <woodruffw> AFAIK, the only way to verify an online identity is to use an out-of-band authentication, like a real-life meetup and key exchange. beyond that, you're relying on degrees of (very reasonable) trust
[23:09:14] <cold_sauce> makes sense
[23:10:40] <woodruffw> yep, which is why PGP uses a tiered trust system instead