Using Networks, Blockchain, and Game Theory To Ensure Privacy on the Web
Quanta Magazine recently interviewed Harry Halpin, an MIT research scientist and CEO of NYM Technologies, about his work on internet privacy. The internet, in its current state, has a large privacy problem. Much of the past cryptography research aimed at maintaining internet security has dealt with encryption, ensuring that nobody can tell what you send to others over the internet. In the interview, Halpin claims that even if a third party cannot tell exactly what one person sends to another, they can still pick up on how the two people communicate. Over the internet, anybody can see who you communicate with, when you communicate with them, and how long those conversations last.
Addressing this privacy concern is difficult, but there are some techniques from this course that we can apply to the situation. The first strategy is something known as a “mixnet,” short for mix network. In a mixnet, individuals are arranged in a network. When they want to send a message to somebody else in the network, they break the message into pieces, encrypt each piece, and randomly send it to an intermediate node, called a “mixnode,” in the network. The mixnode receives messages from many different senders, which all get mixed together. The message goes through several mixnodes before arriving at its destination, where it is decrypted and pieced back together. The network structure and mixing strategy ensure that, at any given mixnode, a third party cannot tell both the message sender and recipient. In this way, a network is used to prevent others from seeing who an individual communicates with.
The second technique from this course that can support the mixnet strategy is blockchain. Mixnets require expensive computation, and there is normally no incentive for somebody to use their computer as a trustworthy mixnode. Halpin suggests that a blockchain can be used to incentivize users to act as a mixnode. Users that run mixnodes are called mixers. Like transactors in the Bitcoin blockchain, users pay a fee to send their messages through the network, and mixers (like miners) do the work to properly mix and send the messages, receiving the fees from their users. Additionally, there are stakeholders that bet on mixnodes that they think will be successful at processing messages. Stakeholders receive part of the reward if they bet on a successful mixnode. In this way, stakeholders are incentivized to bet on relatively unknown nodes because they will not have to share with other stakeholders. As stakeholders discover new, reliable nodes, network traffic is dispersed to those nodes to ease congestion. Using game theory, we can model this process of creating and betting on mixnodes as a game. Diaz et al. showed that this game has a desirable Nash equilibrium, indicating that players cannot game the system.
This article shows how many of this course’s concepts are used to improve privacy on the internet. First, we saw that our idea of a network of computers can be extended to create a mixnet. In this way, a network of computers can send information without worrying that third parties will trace end-to-end communication. Next, we saw that our idea of a blockchain does not necessarily need to be used with a currency like Bitcoin. Specifically, we saw that blockchains can be useful in the transfer of messages, not just currency. Finally, we saw how game theory is being used to prove the feasibility of truly private networks.
Sources:
https://www.quantamagazine.org/the-computer-scientist-who-boosts-privacy-with-entropy-20221018/
Diaz, C., Halpin, H., & Kiayias, A. (2022). Reward Sharing for Mixnets. Cryptoeconomic Systems, 2(1). https://doi.org/10.21428/58320208.00acf460