My Take on Secure Communication

Posted on 2014-03-19

Disclaimer:* I am not really into crypto (yet?) and everything I write here is purely idealistic and probably somewhat unrealistic/stupid

The current situation

Some time ago, Facebook bought Whatsapp, which was my main way of communicating using my phone. This event has made me think about where the internet is heading in terms of privacy protection. Right now, I am trying to figure out whether Chromium could be a better alternative to Firefox, which is working worse and worse on Linux. I purposely do not want to use Chrome as Google already uses their own search engine, YouTube, Adsense, Analytics and Android/the Play Store, and maybe their DNS-servers to track internet users from all around the world. Using their partly proprietary browser which is known for phoning home is obviously not the best idea.

But back to Whatsapp. There are some alternatives, some praising encryption, like Threema or Telegram. Threema is not free and as Google does not utilize any payment method beside credit cards, only few people will actually use it, and the only reason to switch to a messenger is the userbase. I did not like Whatsapp in the first place, but everyone used it, so I had to comply. Now I made the switch to ecrypted (called "secure") Telegram messages with some of my closer friends, but while it claims to be "open", this only means there is an API. And if they sell out one day, say if Facebook or Google offer them some millions, all their promises will be worth nothing.

Another big part of my concerns, before we come to my ideas for a bright future, are email and VoIP. Email right now is de-facto unencrypted, usually transmitted via SSL to the provider but that is it. The providers usually provide for free and most of them probably scan the conversations to sell the information and/or build user profiles (cough Google cough). This can be partially solved by just encrypting/signing emails via PGP, which right now only security nerds and mafia bosses do. But handling keys is just not ready for the masses, it basically should handle keys, en-/decrypt without the average user even noticing. But the decentralized way emails are set up is already quite good, although there are some really big providers, there is no loss in using some small provider you trust or even using your own domain and server.

VoIP is the real problem right now. I primarily use Skype, which is owned by Microsoft since 2011. Skype is known to send all kind of data to Redmond, and also allow crazy amounts of script languages for its ads, which already resulted in some big security holes. Also the Linux client sort of sucks. The interface is okay-ish, but the technical side of things is not. But I do not blame Microsoft for this, I am happy for the existence of a Linux client at all. So, Skype is bad, alternatives are rare. Mumble is nice, but is centred around servers for gaming clans and similar communities, using it as messenger is suboptimal. I also tried Jabber, the VoIP-protocol with many features, which is actually sort of great the way it is, but not end-user friendly enough.


What do I think would be the optimal way to handle this situations and prevent some cyberpunk scenario in which Google, Facebook, Microsoft and of course the NSA know everything about the average internet user?

First, we need to divide. I would keep email, as it is an easy and already established way of exchanging information and data in a mail-like fashion, which we will still need in the future. But we need some way of enabling everyone to encrypt this mail. A simple choice would be to establish a service that handles public keys and pairs them to accounts. You then get a plugin for Outlook or Thunderbird that connects to your account, generates a keypair, uploads the public portion, gets the needed public keys of others and handles en- and decryption of mails. The downside here would be a pretty big single point of failure. If this service is down, or worse, gets hacked, all the security is jeopardized. An idea would be to not store this information on central servers but instead in a Bitcoin-like blockchain which is shared by everyone. Which is probably generating enormous amounts of traffic just to synchronize. Also, if no one uses ad-loaded webinterfaces or provides unencrypted emails, how are the email providers going to survive, and how are they going to justify subsidizing the free email services? People will not pay for privacy, which is the actual root of the problem.

Then we need a new multi-platform messenger, decentralized, encrypted, capable of sending pictures, audio, video and of course live calls including video and conferences. Basically Skype Premium, but without central servers and more in a peer-to-peer fashion. Maybe just define a protocol and write an open source reference implementation of a client. I already thought about this a while, and most of this is accomplishable, Skype already provides most of the functionality across Windows, Mac, Linux and a lot of phones. There is the wonderful Speex-codec which can be used for calls, video streaming should be no problem either. As the content should be completely end-to-end encrypted, compression would be sort of useless, so the generated traffic will be more, but not by a lot.

But here comes the one problem I found rather quickly. You need to be able to sync your private keys across devices. As I propose to just develop a protocol without a central management, you cannot just enter your username and password and get all your stuff from a server, because that would mean, your private keys, which are not supposed to ever leave your devices would be stored on the server. But I have come up with a solution, at least I think so.

There will be some server structure involved, which will hold the private keys, but in an encrypted form. When starting to use the service, you generate your keypair. The public key gets into some sort of public key infrastructure I yet have to figure out, maybe blockchain-like, maybe not. The private key gets stored on the device, protected by a password. A copy of it gets encrypted (say using Rijndael/AES or Twofish) and can be stored on third-party servers. Those third parties might enforce additional security measures via accounts/passwords. If you want to use a second device with the same identity, you can get you private key from said server, decrypt it and import it. This way, you could basically store your encrypted key inside Dropbox. I have to add, that this method relies on strong passwords more than anything else, but maybe this will force sensible password policies into the heads of the masses.

An addition I would make would be some sort of identifier-system, comparable to DNS. No one wants to remember numbers or hashes to identify people, and names are not unique to a person, so there should be some way to (decentralized, of course) link a username or an address with a keypair, which can be resolved easily. So if you want to send me a message, you send it to "sulami", we resolve the name, get the public key and are ready to start an encrypted conversation.

Who will do it?

This is the important question. I would like to work on something like this, but obviously the scope is gigantic and there are experts needed on crypto, network messaging, development for specific platforms, interface designers, and the list goes on. If you read this and are interested in starting such a project, do not hesitate to contact me, someone has to make the first step. My public key is on the contact-page…