Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

MEGAChat now includes end-to-end encryption

mega.nz

Post reply on HN
Print view View original

51–60 of 98 posts

Re: MEGAChat now includes end-to-end encryption

#51

Post by yaur »

c22 wrote: post #47
zanny wrote: post #44

Earlier quoted context omitted.

From a UX perspective, can we do better? I'll base both off my personal experience - Kmail and KDE Telepathy - since those are my email and IM client. In kmail, to use GPG keys, I need to go to Settings -> Configure -> Identity -> Modify -> Cryptography, change my signing and encryption keys (and it supports PGP or MIME) and then, since I don't already have keys, need to open Kleopatra, the KDE GPG client. Kleopatra…

I think before you even tackle the ui issue there's a general knowledge hurdle you have to cross. Most people are unable to reason about or manipulate uis around keys because they fail to understand the purpose and basic mechanism of them. I spent roughly 5 hours explaining the general concepts behind public key cryptography to my girlfriend and this was only possible after making a compelling argument that everyone…

This can't be overstated. Even if we got the general public to sort of understand PGP and PKI there is going to be a significant number of users that don't understand the difference between a password and a password protected key. The expectation is going to be that when they move to a new computer that their "password still works" and are going to be extremely angry when it doesn't.

Re: MEGAChat now includes end-to-end encryption

#52

Post by yaur »

buu700 wrote: post #28

Since no one else has explicitly brought this up: https://tonyarcieri.com/whats-wrong-with-webcrypto (also see Matasano's "JavaScript Cryptography Considered Harmful", etc.). There is nothing that a service like this will protect you from that you don't already get while using TLS in just about every insecure/non-E2EE chat service. If you're looking for actually secure communication that runs as a Web app, my startup…

From your site (attributed to cure53): "Cyph provides security from a broad range of cryptographic attacks"

From the actual conclusion: "Cyph claims to provide security from a broad range of cryptographic attacks"

Whats the deal here?

Re: MEGAChat now includes end-to-end encryption

#53

Post by zanny »

rakoo wrote: post #49
zanny wrote: post #44

Earlier quoted context omitted.

From a UX perspective, can we do better? I'll base both off my personal experience - Kmail and KDE Telepathy - since those are my email and IM client. In kmail, to use GPG keys, I need to go to Settings -> Configure -> Identity -> Modify -> Cryptography, change my signing and encryption keys (and it supports PGP or MIME) and then, since I don't already have keys, need to open Kleopatra, the KDE GPG client. Kleopatra…

I 100% with you that UX is the most important problem that is not solved with PGP, and as long as nobody will do something we'll never be able to use crypto. I also believe that if your crypto doesn't solve something that PGP doesn't already solve, then you'd better spend your time on bettering UX instead. However Mumble is not a valid comparison, because Mumble is not End-to-End encrypted. The security of Mumble amo…

It is end to end encrypted when you are one of the participants and you own the server. Thats what I mean by its the best secure communications solution I have - I can sholder the work of getting a valid cert, but once its setup as long as I trust who I'm talking to is who I think I'm talking to, I am sure nobody can evesdrop because its my server. And for the people I care to have private conservations with, downloading / install Mumble on their phone or desktop is their regular patterns of behavior and isn't inconvenient at all compared to insanity like GPG or OTR.

Re: MEGAChat now includes end-to-end encryption

#54

Post by zanny »

yaur wrote: post #51
c22 wrote: post #47

Earlier quoted context omitted.

I think before you even tackle the ui issue there's a general knowledge hurdle you have to cross. Most people are unable to reason about or manipulate uis around keys because they fail to understand the purpose and basic mechanism of them. I spent roughly 5 hours explaining the general concepts behind public key cryptography to my girlfriend and this was only possible after making a compelling argument that everyone…

This can't be overstated. Even if we got the general public to sort of understand PGP and PKI there is going to be a significant number of users that don't understand the difference between a password and a password protected key. The expectation is going to be that when they move to a new computer that their "password still works" and are going to be extremely angry when it doesn't.

Thats why you cannot expect them to understan PGP and PKI. You want a sane default that can secure 99% of people and expose the guts to the last 1% through options menus.

The perfect PGP secure email client would be cross platform and provide its own email server. The client has included trusted keys to communicate with upstream, but when you first start it you make your account and that generates you a signing and encryption key based off your password, backs it up on their server (remember, its password protected, they cannot access it despite backing it up) and then whenever you sign in on other device your keys are downloaded and just work.

This basically turns PGP into username / password, but the only alternative is to use wifi to send your keys between devices when you first start your client of choice - IE, you start it for the first time, it says "do you already have an account? You say yes and it simply asks which PC on the lan has your keys, it communicates with it, you confirm on the other computer to send the keys, and it synchronizes your keys cross-device peer to peer.

And I worry even that is too complicated for my grandmother to be willing to put up with to stop the NSA from reading her emails.

Re: MEGAChat now includes end-to-end encryption

#55

Post by echochar »

DanBC wrote: post #35
Freak_NL wrote: post #19

It's great that there is such a strong interest in secure communication, but all of the solutions that keep getting placed in the limelight in recent years are either centralised or exclusively available for use with a smartphone (because of the phone number requirement), often both! I'm sure that these solutions are a lot more secure than completely unprotected alternatives, but am I wrong in assuming that if you wa…

> Call me a sceptic, but this is probably why use of these protocols and principals is limited to security experts and software developers who understand these principals OpenPGP is impossible for most email users to get working correctly. Remember that most people don't even know the difference between CC and BCC. Here's an example of de-anonymising anonymous messages, supossedly from clueful users: https://ritter.v…

From ritter.vg:

"You're browsing without Javascript! If you have no idea what that means, you should ask your technical friend about it. Otherwise - kudos."

No, kudos to you. I could count on one hand the number of times I have been congratulated by one of these capability detection messages for browsing without Javascript.

Re: MEGAChat now includes end-to-end encryption

#56

Post by bascule »

junto wrote: post #15
aw3c2 wrote: post #12

Earlier quoted context omitted.

If Kim Dotcom says you can't trust something does that mean you can? https://de.wikipedia.org/wiki/Kim_Dotcom#Werdegang

Nice! For those that don't know, Kim Schmitz / Kimble(now Dotcom), sold out his fellow warez peers to the copyright lawyer Günter Freiherr von Gravenreuth. Like the wild West he was supposedly paid per head. http://www.forbes.com/sites/andygreenberg/2013/04/17/where-k... Not mentioned in the article is his involvement in the AT&T calling card fraud conspiracy (estimated losses $20 million) in the early '90's. It is t…

See also his Security Charlatans page on Attrition.org:

http://attrition.org/errata/charlatan/kimble/

Re: MEGAChat now includes end-to-end encryption

#57

Post by eganist »

yaur wrote: post #52
buu700 wrote: post #28

Since no one else has explicitly brought this up: https://tonyarcieri.com/whats-wrong-with-webcrypto (also see Matasano's "JavaScript Cryptography Considered Harmful", etc.). There is nothing that a service like this will protect you from that you don't already get while using TLS in just about every insecure/non-E2EE chat service. If you're looking for actually secure communication that runs as a Web app, my startup…

From your site (attributed to cure53): "Cyph provides security from a broad range of cryptographic attacks" From the actual conclusion: "Cyph claims to provide security from a broad range of cryptographic attacks" Whats the deal here?

Good catch. Mario actually provided a marketable quote via email (signed and all). Although that quote is intended as a one-sentence summary of Mario's report, it's not strictly a quote out of the report as much as it is strictly a quote from Mario. Therefore, the source on the quote might more accurately be attributed as "Mario Heiderich in summarizing the Cure53 pentest report."

The thinking there was that there wasn't a good way to quote any single part of the report without introducing unnecessary and/or fishy ellipses, because although the point of the rest of the conclusion is that it confirms the claim made by Cyph, there wasn't a readily available standalone sentence in there which could go on the front page.

Background: I advised Cyph on AppSec topics on behalf of Mach37 as a mentor, and I helped coordinate the Cure53 pentest. I'm the one who sourced the quote.

Give me a few moments (or days; we'll see) for me to find a good way to publicly and non-repudiably present the exact source of the quote.

Edit:

Signed statement: http://pastebin.com/HcL5bneg

Key: https://keybase.io/cure53

Hat tip to Mario (https://twitter.com/0x6d6172696f)

Re: MEGAChat now includes end-to-end encryption

#58

Post by garrettr_ »

tptacek wrote: post #48
merpnderp wrote: post #46

How does this compare to something like cryptocat?

It is basically cryptocat.

Have you taken a look at Cyph (mentioned elsewhere in this thread)? Their app is also basically cryptocat, but they've taken the time to engineer an interesting solution [0] to the "Trust on Every Use" problem that secure web applications have.

Of course, you still have the problem of deploying a secure application within the attack surface minefield of a web browser, but that's a separate issue.

[0]: https://docs.google.com/document/d/1PGS50eZB9Ud7wVKmrc0HU_FD...

Re: MEGAChat now includes end-to-end encryption

#59

Post by tptacek »

garrettr_ wrote: post #58
tptacek wrote: post #48

Earlier quoted context omitted.

It is basically cryptocat.

Have you taken a look at Cyph (mentioned elsewhere in this thread)? Their app is also basically cryptocat, but they've taken the time to engineer an interesting solution [0] to the "Trust on Every Use" problem that secure web applications have. Of course, you still have the problem of deploying a secure application within the attack surface minefield of a web browser, but that's a separate issue. [0]: https://docs.go…

Secure messaging has to work. All the time. If your threat model is criminals, Gchat will protect you just fine. You use encrypted messaging to prevent governments from reading your messages. Look at the Telegram/Iran fiasco for an example of this.

So: my recommendation is, never use any secure messaging system that boots off a web page. No matter what they claim to do to mitigate the risk.

I'm actually, for a bunch of specific reasons, not super comfortable with browser extension applications either. But at least the browser tries to protect an extension you have to explicitly install and explicitly run.

Re: MEGAChat now includes end-to-end encryption

#60

Post by doomrobo »

Somewhat off topic, but does anybody remember when MEGASync (a MEGA desktop sync client) came out and the Linux source code was "on its way?" It's been over a year (maybe 2?) and I've seen nothing. I have no reason to trust the security of the file storage backend until then.
Post reply on HN