What are you working on? Checking in with Daniel Honerkamp
Hi, I’m Daniel. With a background in both machine learning and economics, two of the domains Fetch.AI is working in, I couldn’t be more excited to see an economy of Autonomous Economic Agents (AEAs) emerge. At Fetch.AI, I’m currently working on our consensus protocol, the blockchain layer upon which this new, decentralised world will operate.
In this article I will be focussing on the role of fees and economic incentives in a decentralised network. In contrast to centralised systems where fees pay for security measures, in most decentralised systems, mining rewards essentially are the security. Let me elaborate. Consensus is all about agreeing on the state of the chain and who can alter it. To enable both open and anonymous participation in the consensus, blockchain systems have replaced the classical mechanisms of centrally controlled access and reputation with the expenditure of a costly resource (hashrate in Proof of Work, locked up stake in Proof of Stake). As a consequence, the ledger’s security directly depends on the total effort put into the chain, which in turn will not exceed what can be earned from doing so. Fees and mining rewards are thus at the very heart of today’s decentralised systems. This relationship becomes less direct when we consider new developments such as finalisation, or the possibility to revert attacks. But the underlying principle remains relatively unchanged.
We have a large number of different actors participating in the consensus, taking on a variety of roles, and the interests of each actor should be aligned with the protocol rules and total welfare. While everyone likes to talk about honest nodes, in the end we can only rely on rational actors that act to maximise their own utility. This understanding informs every design decision we make and it is for this reason we seek to reduce the influence of any single participant, as we strive to create a minimal agency consensus.
Given the necessity of mining rewards, there are two main ways to levy them: seigniorage (newly minted tokens) and transaction fees. The central challenge here stems from what economists refer to as externalities: each transaction influences system security, processing costs, bandwidth and storage of all other (and in some cases even future) users and miners. While optimal pricing would internalise such costs and benefits imposed on others, this often requires knowledge of unobservable quantities that are hard to estimate. At the same time we want to avoid over or underpaying for the system’s security. These are not completely new problems. Indeed, extensive literature on pricing and auction mechanisms already exists. The difficulty is to adapt and employ them within the constraints of an anonymous, decentralised system. What we find is that, once again, we can benefit hugely by minimising the agency of certain actors and this greatly expands our options.
Finally, we have to test the implications of our choices. To do so, we draw heavily on the tools of Game Theory and simulations.
What makes the work at Fetch.AI both challenging and rewarding is the new, unexplored interplay of Computer Science, Economics and Game Theory. This is all conducted with the goal of designing a decentralised system, operating at high throughput, within an existing world of outside options, at the lowest possible cost. While this creates tough problems to overcome, the immense progress that has been made in the decade since Satoshi Nakamoto first introduced us to Bitcoin shows what can be achieved. I am very excited about the fantastic concepts and implementations we are building here at Fetch.AI.
