Thoughts on crypted communication

Due to the recently published information about mass surveillance on a yet not known level right now the question remains how to encrypt communication. I had some thoughts regarding that topic involving a GPG like web of trust combined with user friendliness which I’d like to share here.

Given everything which was published so far, un-encrypted communication is not save at all. The same is unfortunately true for encryption methods which rely on encryption provided by the servers of some organization. If there is a centralized organization storing the keys for you, or just providing you with the encryption technology, you are screwed, because the intelligence agencies will force them to cooperate. For that reason, the encryption must take place on the end users system already (and the software must be Open Source).

However, if you have end-user encryption, you have the problem of the key exchange – if two people want to communicate securely, they need to exchange the keys or at least securely verify that a public key indeed belongs to their private key. That only works if they meet in person – or if there is a web of trust.

A common example of such a web of trust is the GPG web of trust: people who have properly verified that person A belongs to key M sign this key. If person B trusts person A, it can just use key M since it is already verified by person A. However, in case of GPG the web of trust never reached mainstream. It is mainly used by technical minded people. Most users never got used to it.

So, from what I can tell the only chance to establish a web of trust is to hide the technical details as much as possible from the users. The same is true for the actual key exchange – it needs to be as simple as possible so that each normal user can use it.

Given this background I would suggest the following solution at least for mobile phones. You download the app, and it asks the user for a password. In the background, a key pair is generated and secured with the password, and all data stored on the device are encrypted using the public key. If user A meets user B all they need to do is pressing a button in the app, and a QR code is shown. The other user scans that QR code, and its done. The user shows up in the contact list, and they can chat. In the background, the app extracts the key ID and fingerprint from the QR code, downloads the public key, signs it and uploads the signature automatically.

The biggest problem comes up when user C comes into play, wants to communicate with user A, but they both have no common connection in their web of trust. They would have to meet – or use some other way of exchanging the data securely. A simple way would still be to talk on the phone, but that never worked for GPG. So some kind of web service to host their QR code for a short time only would probably a solution, although it would be pretty risky.

To lower the danger of a man in the middle attack in the above given web example the key servers must only accept one key pair for each identity, which is different to the way GPG works. That would in fact mean that you can have each login only once – if you loose your key, your are screwed.

One question though remains: how many steps in the web of trust are still trustworthy? I guess that could be left as a configuration option if, and only if, a user wants to modify that.

To summarize: I guess that the current cryptography technologies we have could really help to establish secure communication. But to really bring that communication to the masses we need easy-to-use (read: your grandma!) applications doing everything in the background.