Privacy preserving engineering

[gpestana]

One thing that is interesting about P2P and decentralised networks is that they tend to the "property extreme" (!lacking of a better name..). Consider, for example, scalability. P2P networks have better potential to scale compared to centralised (and distributed) client-server systems. That potential comes from the fact that network users can contribute with resources for the network to function, resulting on very interesting and self-sustained ecosystem in which scalability is elastic and self-regulated. The problem is how complex it is for designers and developers to turn that potential into reality: P2P protocols and networks are hard to design and to implement. It's hard to get all the pieces right: from low level networking to peer interaction protocols, incentives design, security, trust, privacy (and the list goes on..). Potential and complexity are the forces behind the "property extreme" in P2P decentralised systems: a system can potentially provide very good scalability but if it is poorly designed and implemented scalability will end up being a bottleneck instead of an edge.

Same goes for privacy. One of the most interesting propositions of P2P and decentralised networks is its potential for respecting user's privacy by design. Privacy by design becomes even more promising when compared to the current state of affairs in which the services we use online are owned by companies with the means and the incentives to register everything about their users. In P2P and decentralised networks, there is no central entity which sieves through all what happens in the network. However, P2P designs often require many to many communication patterns and collaboration. Users in the network often work as message relayers (one of the ways for everyone to contribute with resources to the network) and need to make decisions based on their partial view of the state of the network. With such constraints, network users (or peers) need to collaborate and learn about each other's partial view of the network state. At a higher level, these interactions often result on a lot of systemic leakage of privacy-sensitive information: what a peers knows over time may disclose what she's been up to. Or correlating the message patterns over time of a subset of network peers may disclose what data they are consuming or serving. This information leakage is even more important because it is disclosed not to one entity as in centralised systems, but potentially to the whole network. Again, the "property extreme" is at play: there is a lot of potential to design and implement systems with the best privacy properties, but it will most likely be leaking private data all over the place.

So there is a problem: it is hard to design and implement P2P and decentralised networks and ensure the correct properties. For privacy preserving networks, there is lack of primitives and implemented protocols that can be easily used and adapted by developers. There is also a rather large chasm between academic research on anonymous communications and privacy preserving networking and the mechanisms and protocols implemented by current P2P projects. I argue that if we want to design and build privacy preserving networks, we need as a community to invest more time on:

• building modular and easy to use primitives and protocols for privacy preserving P2P networks;
• improve developer and system designer awareness about privacy enhancing technologies;
• close the gap between academia and industry: bring the of the art on research of privacy preserving networks to the industry as soon as possible;
• measure and assess privacy in current P2P networks and understand how to improve it .

Privacy preserving engineering is not a thing. Yet. We at hashmatter are working on projects that enable developers and P2P system designers to build privacy preserving networks. We're building p3lib, is a library with modular privacy preserving primitives and protocols (p3) for routing and messaging in P2P networks. We've also started PPP, a list of primitives and protocols that can be used in networked systems for improving privacy and anonymity. And we've been also researching on privacy preserving networks. We also believe that the community could use more network security and privacy audits of P2P networks, so we are also putting those efforts together.