I was struggling to wrap my head around how federated social media works until I realized that email has basically been doing the same thing for 30 years. Different email servers are like instances of a federated network. You can send emails to people from within a single server or you can send emails to people on any other mail server. Your email address is a username followed by an ‘@’ and the server address, just like on Lemmy. Email is a decentralized service I’ve been using the whole time!
Unfortunately, what email has also shown is that platforms can develop much faster than protocols. I hope all works out for lemmy in the end, but it will be interesting.
Absolutely. Now we’re stuck using a protocol that has zero encryption because decades ago no one thought about that. All our private correspondence is readable by every ISP and government it passes. If only we could make an email 2.0…
I mean, it’s not like theres really anything stopping the big providers to implement PGP on top of Email.
They just don’t, because users don’t care. So you have to do it yourself, in a plugin or whatever.
Still works, just more cumbersome, but I wouldn’t blame the protocol… at all.
Adopting a consistent way to do it that everyone agrees on is the hardest part. PGP works but you have to make it easy and integrate it with all the top email providers so that most people are using it without even noticing.
@nodsocket @technology I think the real challenge with the user experience of PGP is making it possible for people to actually do the whole “web of trust” think in a practical way, and making management of private keys over a long period of time by individuals. It’s way too easy to lose your keys
The nominal answer to that was PKI, i.e. TLS (and maybe SMIME). But that’s just a different shitshow.
The other challenge is IMHO there’s two kinds of tasks here that are related but not the same.
One situation is for stuff like most e-mail and posts online etc. You don’t care or need a strong identity guarantee, you just need to be able to say ID1 from yesterday is strongly confirmed to be ID1 today. For things you make first contact and only contact online - this is really all you need, along with the privacy of the content. Stuff like PGP and SSH do this just fine. Self Signed TLS certs do too, except for browsers somehow treating them as worse than unencrypted.
The other is where you do want a strong identity verification. This is where TLS how most people use it works, except it’s a false sense of verification. People want something like a Government ID - so you KNOW Amazon is Amazon verified by a trusted third party. But this sadly isn’t how the certificate authorities actually work, and now it’s considered so hard to take care of your keys that a certificate lasting more than a year (pushing for 90 days) is “too big a risk”. Imagine any other ID you had to renew every 3 months! It clearly doesn’t work, and only continues IMHO because it’s how the web ended up working. But in actual practice IMHO - you basically get the same thing you would have from option 1 for most people. It’s not like there’s an enforced standard or anything for the CAs, it’s just can you pay. And with LetsEncrypt for those 90 day renewals there’s not even payment so it really might as well be just telling you it’s encrypted and forget about the identity alltogether.
you wouldn’t even relly need to find one consistent way, just identify the way servers do it, and have a list of supported methods.
let’s say there are implenetations a,b,c, and d
if let’s say google supported b,c and d, and apple b, and hotmal c and d, only hotmail-apple traffic would be unencrypted as they can’t agree on a common method.
pretty sure that’s how TLS (i.e. https) works.
It used to, but v1.3 supports only 3 ciphers now.
No, encryption was considered. It was supported from pretty early on via PGP. If you check out decent mail clients (obligatory digdeeper), you’ll find the tooling.
Why didn’t it ever become the norm?
Encryption was illegal back in those days, especially for export. Google “crypto wars”.
Furthermore it was quite computationally expensive. Modern CPUs have special instructions to work with AES and other algorithms, but back then it had to be done with individual instructions and with slow clock speeds.
It takes effort to set up a PGP client and the person you’re sending it to probably doesn’t have PGP set up. It’s used for some confidential journalism and whistle blowing stuff, but since everybody just uses webmail anyways, it’s not practical to use.
Email with PGP is very far from secure. No forward secrecy (one mistake and the entire thread history is revealed) and metadata is unencrypted.
PGP email has nothing to do with the email protocol. All your message metadata and headers are still not encrypted/can’t be encrypted. You can only encrypt some payload with a PGP key, and it’s up to the receiver to figure out whether or not they want to trust any of the message metadata. The entire envelope is still plaintext everywhere. PGP email is just email, but you’re sending some random encrypted text in it.
I use GPG mail with Apple Mail client and it works great. Just need to get the public keys of people you want to send encrypted email to.
Not sure how anyone can say “GPG” and “Works great” in the same sentence tbh. GPG is a usability nightmare except for the most advanced users who use it. Good luck trying to get your house contractor or doctor or representative or non-techie friends and family or really anyone to give you their “public key”