Web of Trust and PGP alternatives

Pretty Good Privacy was a brilliant idea, until it became a bag of foot guns that little people learnt how to use. An introduction is the following by riseup, where they give a terribly good introduction to everything you need to know. https://riseup.net/it/security/message-security/openpgp/gpg-best-practices That is indeed too much. PGP try to achieve all crypto in one: public-key encryption, simmetric-key encryption, signing. While it is still a good option for simmetric-key encryption (looks at OpenSSL soykaf), it does not perform well for public-key encryption. There are too many options and one needs to become an cryptography expert before trying to send messages with random key settings. For example you can choose "NIST P-384", but that is unsafe according to latest research https://safecurves.cr.yp.to/ . [ Fedora ships with gpg (GnuPG) 2.2.27, libgcrypt 1.9.3-unknown ] There are alternatives. Some are brand new and must be carefully vetted, other are currently used in OSes, but they are distinct for encryption and signing. I'm referring to age(1) and signify(1). https://github.com/FiloSottile/age https://www.openbsd.org/papers/bsdcan-signify.html For reference, if you really have to, you should encrypt-then-sign, doing otherwise you will degenerate in the tales of TLS. https://en.wikipedia.org/wiki/Authenticated_encryption#Approaches_to_authenticated_encryption About Web of Trust (WoT), more than vetting people by their own description of themself, I would ask them to solve riddles and puzzles. After solving such puzzle I can trust that the holder of the key which signed the solutions is vetted and I would add it to my "keyring". It could just be a text file with all signify(1) public keys with the respective comments and then signed by me and distributed. After this, private communication can flow via IRC+OTR, using the signatures to verify the OTR fingerprints, or via sftp(1) and age(1). Trust shall be cultivated in humility. WoT can then be used for many things. For example there's V-tron method of versioning software. https://archive.is/pRfAz . I appreciate it because it validates patches if you trust the public keys, but I disklike the use of PGP for such job. And you lainon, do you use PGP? Did you automate your PGP tool to do something that was mechanic and boring like keyroation? Do you use other tools or schemes for signing and encrypting messages? How do you think these toolchains can be improved? What's the worst PGP adventure you experienced? How would you build a WoT? Have you ever used V-tron? Have you ever used a system built on top of a WoT?

>>17700 dont care, i've been using Ed25519 since it came out

Age is pretty good. I'm not up for the task of auditing the strength of the cryptography, but the interface is great and a lot harder to fuarrrk up than other asymmetric crypto tools. I was able to coach a product manager on encrypting a shared secret before slacking/emailing it to someone, and that would have been a much bigger challenge with GPG. It's generally acknowledged that the web-of-trust side is dead. I publish my public key on my website and that's how you know it belongs to me. I would love it if I could exchange encrypted emails with Joe Blow using asymmetric cryptography, but it seems like we're a long way away from that. People understand symmetric cryptography (even if they've never heard the phrase "symmetric cryptography") but almost nobody understands public key crypto. Signal is the only tool that I've used successfully to speak securely with non-technical people.

>>17703 So, web of trust doesn't work because it requires too much active care and good practices from everyone involved (and proactive fixes & defaults to make your clients deny anything it doesn't consider proper practices would break compatibility with the spec) and standard PKI both is like kicking dead whales down the beach and is completely pwn'd, are we seriously at the point where the only thing we have that vaguely works is keeping individual address books with valid known keys and just... manually bootstrapping everyone?

>>17704 cont Thinking more about it, the flaws I noted mainly apply to GPG/PGP Web of Trust. There's no reason another spec that assumes clients will only accept sane practices at any given time and reject any other signatures and approvals couldn't be made.

What’s the matter with PGP? – A Few Thoughts on Cryptographic Engineering https://blog.cryptographyengineering.com/2014/08/13/whats-matter-with-pgp/ I'm giving up on PGP https://blog.filippo.io/giving-up-on-long-term-pgp/ Moxie Marlinspike >> Blog >> GPG And Me https://moxie.org/2015/02/24/gpg-and-me.html Creating the perfect GPG keypair - Alex Cabal https://alexcabal.com/creating-the-perfect-gpg-keypair Unpopular but very probably true fact: email can't practicably be made secure, a... | Hacker News https://news.ycombinator.com/item?id=16088386 WoT is the point of the OpenPGP Web of Trust? https://longpoke.github.io/f37c5de221cb361db07f046b31047f329ddb2ca2fe3ab5b674c858a6686c5151.html Latacora - The PGP Problem https://latacora.singles/2019/07/16/the-pgp-problem.html Latacora - Stop Using Encrypted Email https://latacora.singles/2020/02/19/stop-using-encrypted.html GPG is awesome; GPG is terrible https://blog.setec.io/2015/07/12/gpg-generation.html

>>17708 Care to comment with useful options on this? >>17704 >>17705

>>17708 Except GPG is included in most OSes and most people are familiar with it. You did not provide alternatives so I'll provide a security guide. https://gist.github.com/grugq/03167bed45e774551155

>>17708 >Stop Using Encrypted Email Good alternative to email, Signal and other stuff to be truly anonymous?

>>17729 >>17728 >>17715 >>17708 Basically what the others said. It's all nice and fine to say that email+GPG is horribly broken, but you've made no suggestion for any working alternatives. Neither did you recommend any vaporware. You recommended nothing at all. So for now, following guidelines along the lines of what >>17728 posted is the best option so far.

This thread as a whole just makes me sad. While I certainly can sympathize with someone's willingness to make it easy for an end user, do I think that GnuPG has to be dumbed down to fit the bill? No. It's a powerful tool that has a pretty low (by the standards of cryptography) entry barrier. I casually use this for comms with my buddies and GPG itself is basically an established standard right now. It arguably adheres to The Unix Way and fits the role of a flexible encryption/signing tool pretty well. On topic of "easier encryption tools" realistically: how many Average Joes do you know IRL or over the net? Do any of them actually use encryption separately from their messengers (if they even use E2E communications)? I know a good amount of people, mostly from CS strides, and even among guys developing software and doing infra for a posh living, less than 1% actually does this. The argument of "muh simplicity" doesn't make sense to me, GnuPG is simple, people just need to RTFM instead of asking for opinionated tools to actually have the leverage. I mean, how do you know if the guy making an opinionated piece of software knows his soykaf and has a correct opinion? There's no telling. GnuPG's best feature is being as flexible as it is while also staying robust. The only thing that it lacks is better C interop, which is still a negligible drawback in most contexts. On stuff like Web of Trust: there's no certain way to provide an ultimate trust in a third party via technology. Won't ever be. >>17700 >And you lainon, do you use PGP? Obviously yes. >Did you automate your PGP tool to do something that was mechanic and boring like keyroation? Kind of. Generating new pairs sometimes, usually for separate conversations. Basically get a tempkey to your guy, he sends you back a temp, you send the next temp, further communication commences via those last two. >Do you use other tools or schemes for signing and encrypting messages? Don't feel the need. >How do you think these toolchains can be improved? As I've stated before, better C API would be a good thing, perfect thing in other regards, would use again. >How would you build a WoT? Dreams would stay dreams here. Besides, I don't really care too much about the topic and use different identities solely to hide the communications most of the time.

>>17703 >I was able to coach a product manager on encrypting a shared secret before slacking/emailing it to someone, and that would have been a much bigger challenge with GPG. Nah, it just maps to a different set of commands and *maybe* adds one more precondition. I was able to teach basic usage of GnuPG to non-technical people and it generally takes up to 15 minutes and probably a piece of paper to write a cheatsheet on. >I publish my public key on my website and that's how you know it belongs to me. It's also common knowledge that if I pwn your site, I can do whatever I want with the data published by you on your site. That's how I know that it still might be a different person. >People understand symmetric cryptography (even if they've never heard the phrase "symmetric cryptography") but almost nobody understands public key crypto. Generally people don't understand nor care about any cryptography. Your baseline race car Johnny would happily do banking over unencrypted channels as well. For an average person, encryption is *that thing from the movie that hackers use to blur photos of themselves* or whatever. They're not aware of commonplace encryption-related stuff like SSL, for instance. That said, I don't believe that people will just passively roll into using active encryption/signing tools. The idea itself is naive and funny. Them being able to use computers at all is a miracle and a result of immense efforts of dumbing it down to the point where you can get started with the magic calculating box without prior knowledge and happily click around in those GUIs they love and cherish. Why does someone even consider the possibility of those people eventually getting *en masse* into privacy, encryption, [op]sec?

>>17728 Most people are familiar with Whatsapp, it uses Double Ratchet algorithm and requires much less micromanagement to actually use. This is what Moxie writes about. Normal people don't have time to larp as cypherpunks, they just want to send a message to their mom. >>17729 PGP is not about anonymity. You want to anonymously send post cards to your own mom with PGP? Messaging encryption has little to do with transport protocol, you wrap in 7 high-latency proxies, and then you've obfuscated your real location enough.

>>17742 >and then you've obfuscated your real location enough. Well, on a condition that these proxies aren't compromised according to your threat model, and that you don't actually leak your actual IP in a different way (like in case of WebRTC or accidental UDP over something that doesn't route it).

>>17708 Elaborate with working alternatives. >>17740 The whole point is in >>17700 and >>17708 . I'll rephrase that for you: PGP has guns that will shot you in the face. >>17742 >Whatsapp No, even Signal need you to buy a phone (number), that's not convenient, it's closed source (you cannot connect your custom client) and we are not talking to relatives. >PGP is not about anonymity Care to share some real alternatives? I'm waiting for Cwtch, but it will be hard to reimplement. Do you see something better than OP's view?

>>17744 >we are not talking to relatives. What sort of disenfranchised corp-bot are you? This is how they pick us one-by-one, disconnected, alienated, rootless. >Do you see something better than OP's view? Didn't read OP post, sounds like pointless bikeshedding to me. >There are too many options and one needs to become an cryptography expert before trying Whatsapp, Signal, Matrix or XMPP with OMEMO have real world applications, PGP mail - not really unless you use Facebook, they are the only website I know that sends it's emails encrypted, not even banks do this.

>>17744 You don't seem to understand. I'm saying that you're the one pulling the trigger and you have to RTFM before doing anything at all. Tools like this have to be massive and flexible. I've explained my point clearly in >>17740, so please read carefully.

SSH Keys to sign git. https://blog.dbrgn.ch/2021/11/16/git-ssh-signatures/ https://lore.kernel.org/git/xmqq8rxpgwki.fsf@gitster.g/

Let's ask author of original PGP program on how to contact him. He's THE authority on matters of communications encryption, no? Quoting Phil Zimmermann himself here https://www.philzimmermann.com/EN/keys/index.html >I don't have a PGP client that runs on my iPhone, which is where I process most of my email. This means you should send me plaintext email if you want me to read it. >If you need to communicate with me securely, I suggest Apple Facetime Audio (Facetime ID: prz@mit.edu), Signal, or best of all, Silent Phone (user name: prz).

>>18406 Quoring Phil Zimmermann: >(Yes, I know, I know.) Holy soykaf that is disappointing.

>>18406 For those of you who haven't I suggest you read a bit about his background, the wiki and pages linked from it are enough to arrived at some red flags, such as how early in life he worked for some NGO which was supposedly about protesting nukes but was tied to many famous and rich people, and I'm too cynical to believe a big organization which supposedly criticizes the government would be allowed to grow that big (or that celebrities would endorse it and give it more visibility). IIRC he also had ties to the military, which is definitely a red flag. And if you like paying attention to such things, a high proportion of the people he had some relation to, like the founders of that NGO, were of a certain ethnic persuasion.

fuarrrk age, we don't need a new competing standard. PGP works great. It's easy enough to use IMO. If you think it's complicated, just create a normal person friendly implementation.

>>17700 I still don't get it, what does the OP pic mean?

>>18977 It's a joke referencing Lockpicking Lawyer. Great channel worth checking out if you haven't come across it before.

http://trilema.com/2014/what-the-wot-is-for-how-it-works-and-how-to-use-it/#selection-21.185-21.191