Live data from Hacker News

Improved Authentication for Email Encryption and Security

protonmail.com

11–20 of 77 posts

Re: Improved Authentication for Email Encryption and Security

#11
This all seems to be a web-based application (https://github.com/ProtonMail/WebClient). How are the security issues regarding knowing that you're always running that code and that the server isn't compromised and sending altered code? The arguments against server-supplied, js-in-the-browser crypto have been done to death.

Why is this any different, and why am I wrong to dismiss it out-of-hand as (in)secure as simply sending unencrypted data to the server? Why isn't this only an open-source, native app (where I can load a specific, known version instead of whatever is on the server).

> we choose our own primes rather than those used by TLS

Does TLS specify any primes? You can use your own DH primes, SRP primes, and your key is your own prime. Those RFCs recommend primes, but allow the server to use different ones. TLS, SRP, or DH doesn't "use" a single prime, any prime satisfying the requirements in the RFC is acceptable. know it's nitpicking but something about how it was said rubbed me the wrong way.

I would love to know how they communicate between their TLS-SRP layer and their authentication layer. Most implementations are file-based. Did they write a plugin for gnutls or openssl? Did they write their own TLS layer?

I would love for TLS-SRP to be more wide-spread, but this is always the biggest hurdle to adoption in my case.

Re: Improved Authentication for Email Encryption and Security

#12
post #3

how practical is it to drop GMail for these guys? I'm tied fairly heavily to the Google ecosystem (Chome, Play, Finance, etc etc). They already have a mountain of data on me, but I really want to start taking encryption and privacy more seriously.

One downside what totally prevents me from even giving ProtonMail a try is vendor lock-up. You'll be stuck with them with no option to migrate away while keeping historical data. No IMAP, no POP3, not even a proprietary backup format, nothing at all.

One could probably automate a browser to log in, iterate mailbox and dump contents, but that certainly won't be fun.

(Also, no way to use your own pre-existing key, only somehow generating a new one somewhere - not sure how exactly it's implemented and whenever I can trust it all...)

Re: Improved Authentication for Email Encryption and Security

#13

This all seems to be a web-based application ( https://github.com/ProtonMail/WebClient ). How are the security issues regarding knowing that you're always running that code and that the server isn't compromised and sending altered code? The arguments against server-supplied, js-in-the-browser crypto have been done to death. Why is this any different, and why am I wrong to dismiss it out-of-hand as (in)secure as simpl…

>Why is this any different

Nobody says it's different

>Why isn't this only an open-source, native app (where I can load a specific, known version instead of whatever is on the server).

OK, let's suppose you're using a native app. One day vendor issues an update with some critical vulnerability patched. Unfortunately, another vulnerability (or even backdoor) sneaks into this update for whatever reasons. How is this any different?

Re: Improved Authentication for Email Encryption and Security

#14
post #5
post #3

how practical is it to drop GMail for these guys? I'm tied fairly heavily to the Google ecosystem (Chome, Play, Finance, etc etc). They already have a mountain of data on me, but I really want to start taking encryption and privacy more seriously.

Currently, for me, lack of a calendar is holding me back. As well as some other features, but I could probably overcome those. Relevant uservoice: https://protonmail.uservoice.com/forums/284483-feedback/sugg...

Yeah, but it's planned. They also mentioned the calendar coming in a recent blog post.

In general, I should say I am pretty happy with protonmail. I've been using it for over a year now, however, in general I'm still pretty dependent on google for calendar, android, etc.

When it comes to spam, my feeling is that I do get more now. But as long as it is properly categorized -- which it is -- I don't really mind getting it.

Re: Improved Authentication for Email Encryption and Security

#16

This all seems to be a web-based application ( https://github.com/ProtonMail/WebClient ). How are the security issues regarding knowing that you're always running that code and that the server isn't compromised and sending altered code? The arguments against server-supplied, js-in-the-browser crypto have been done to death. Why is this any different, and why am I wrong to dismiss it out-of-hand as (in)secure as simpl…

>How are the security issues regarding knowing that you're always running that code and that the server isn't compromised and sending altered code?

A unique solution might be to use service workers to handle content "updates". The assets could be cached locally, and a single network request could be fired to check for updates. If one exists, the user is prompted to update. This would at least alert you to review the code being sent to your browser if you're so inclined.

Ultimately though it comes down to yet another tradeoff of security vs convenience. Such a feature might be better off as an opt-in extra security feature.

Re: Improved Authentication for Email Encryption and Security

#17
>In ProtonMail’s one-password mode, the mailbox password is derived from the login password via a one-way cryptographic password hash.

I wonder what the password change procedure will be when you have several gigabytes of mail in your mailbox? Would you have to download every message, re-encrypt it in your browser and send back?

Re: Improved Authentication for Email Encryption and Security

#18
post #4

So how does one migrate from the two password to the one? I like the idea of protonmail, but since they made it incompatible with normal public key encrypted mail it's pretty useless for many of us, unfortunately...

PGP is quite difficult to use by most people, and it doesn't even support forward secrecy, which is a huge weakness. It will never be used by more than a core group of highly technical, which is maybe less than 0.01% of the population.

If we're to push end-to-end encryption to the masses, then we ought to try to get forward secrecy in it, and it should be quite invisible to the user.

That's not to say that ProtonMail is getting it right, but it's at least one of the few that are striving to move in that direction.

Relevant post from Moxie from Open Whisper Systems:

https://moxie.org/blog/gpg-and-me/

Re: Improved Authentication for Email Encryption and Security

#19

>In ProtonMail’s one-password mode, the mailbox password is derived from the login password via a one-way cryptographic password hash. I wonder what the password change procedure will be when you have several gigabytes of mail in your mailbox? Would you have to download every message, re-encrypt it in your browser and send back?

Hopefully the mailbox encryption key is actually static, and they only store it encrypted with a key derived from your login password. If you use two different salts the knowledge of the login pw hash does not yield the mailbox key encryption key.

Re: Improved Authentication for Email Encryption and Security

#20
post #3

how practical is it to drop GMail for these guys? I'm tied fairly heavily to the Google ecosystem (Chome, Play, Finance, etc etc). They already have a mountain of data on me, but I really want to start taking encryption and privacy more seriously.

I didn't miss anything when I left Gmail, but I imagine it depends a lot on how you use email. I switched to Fastmail on my own domain and actually prefer the webmail interface over Gmail. I mainly use IMAP via Mail.app on my laptop and phone, Pantheon Mail (formerly Geary) on my desktop. I use PGP wherever I can. I haven't received any spam at all yet, so I can't comment on how their spam filters compare to Gmail. I…

Right now, I also try to keep Google as far away from me as I can. Unfortunately, I don't own a domain myself. Have you tried out ProtonMail yourself? If so, do you think that it is a good alternative to a self-hosted server like Fastmail?
Post reply on HN