Hi, I’m Jenny. I am a theoretical physicist who recently transitioned from the fascinating, but abstract, world of string theory research into the exciting blockchain and AI industry. After completing my PhD in string theory I worked as a postdoctoral research fellow for the University of Tokyo before joining Fetch three months ago as a research scientist.
At Fetch I work in an interdisciplinary team (with a surprising number of physicists!) whose task is to design the consensus protocol for the Fetch ledger. This is the procedure by which decentralised trustless networks reach agreement. The consensus protocol is a core component of the distributed ledger. It plays an important role, ensuring the security of the system and maintaining the efficiency of the network. It is also the backbone of the decentralised framework, in which Fetch’s Autonomous Economic Agents (AEAs) conduct transactions.
The design of the consensus protocol has been shaped by the desire to analyse economic incentives for the participants in the protocol and to develop methods to secure the ecosystem against malicious entities. In particular, Fetch is aiming to combat the possibility of censorship through the use of a Directed Acyclic Graph (DAG), without sacrificing performance on transaction throughput. Part of my work involves simulating the different strategies that nodes may employ to construct the DAG and the impacts that they have on the efficiency and stability of the protocol. By doing this, we can identify undesirable behaviour and introduce safeguards to eliminate or mitigate the effects of such behaviour. More detailed information on this topic can be found in our yellow paper and in previous posts by my colleagues Marcin Abram and David Galindo.
A nested stochastic block model transaction network simulated with three resource lanes
I am also involved in modelling and simulating the economic incentives employed in the design of Fetch’s smart wallets. The use of resource lanes, also known as shards, and smart wallets aims to maximise throughput by encouraging data clusters to dynamically self-organise onto specific resource lanes to minimise congestion on the network. In order to do this, it is vital to utilise the correct pricing model for transaction fees. Complex mathematical modelling and simulations are necessary to accurately determine the resulting effects on the transaction network.
The difficulty in the tasks described above lies in trying to model, predict, and possibly safeguard against the actions of our end users. It is both incredibly interesting and, at times, frustrating (in a good way!). Coming from a background where I worked in the realm of the highly theoretical, it is this aspect of work that I find most rewarding and I cannot wait for the network to go live.