What are you working on? Checking in with Marcin Abram

I am a theoretical physicist and a computer scientist. At Fetch.AI, I work as a Lead Research Scientist in a group tasked with designing a novel consensus protocol for Fetch’s distributed ledger. It is a challenging mission, both because this field of the research is very young and due to the field’s interdisciplinary character. My research draws on knowledge from such fields as economics, cryptography, as well as the social and computer sciences. We try to answer questions about the security of the blockchains. In particular, we model the incentives of participants to ensure the long-time stability of blockchain system solutions. We also prepare code of cryptographic modules and investigate how to make ledgers more scalable without compromising a security of the entire system.

In recent years, many consensus protocols¹ for blockchain have been proposed, both in academia and in the private sector. However, for most of those protocols, the required cost of the consensus scales super-linearly with the adoption of a particular blockchain. This imposes severe limits on how important or popular those solutions can ever become [cf. Budish 2018, Abadi 2018]. At Fetch, we investigate ways to overcome these limits. One idea we try to investigate is to create a protocol where (without compromising the system decentralization) miners have dynamic incentive to follow the rules. This would qualitatively change the relation between the security level and its cost, making the consensus much cheaper and scalable [cf. Abadi 2018]. Another idea would be to create leaderless protocol, where blocks are created truly in a decentralized fashion, making it harder for some attacks to be executed.

Despite the fact that the research I’m conducting is outside the domain of theoretical physics, I find my academic experience extremely relevant to my work at Fetch. My PhD explored computational methods that can be used to solve certain classes of models describing magnetism and superconductivity in strongly-correlated quantum systems. Scientists often apply reductionist approaches to problems they examine. That paradigm doesn’t however capture some emergent behavior in systems where many elements interact with each other [cf. “More is different” by P. W. Anderson 1972]. In physics, one of such example would be superconductivity that rises as an emergent quality, something more than just a sum of the individual elements of the system. During my PhD I was lucky to have a great mentor, Prof. Józef Spałek (Jagiellonian University, Krakow, Poland), who, like Anderson, always acknowledged emergence as central to every theory that describes complex systems [cf. Spałek 2013]. At Fetch, to ensure the blockchain security or stability, it is not enough to look at individual elements of a blockchain problem in isolation. A more holistic approach is needed. One could design the best consensus protocol, but if the incentives are not aligned correctly, it might happen that each rational participant could be incentivised to behave dishonestly. In such cases, the majority of users would quickly deviate from the protocol and the system would eventually collapse.

This is only a snapshot of problems that we try to attack at Fetch.AI, and all of them are novel, difficult, and therefore, exciting. We attend conferences, we publish, and we try also build a final product that would be deployed and useful in the real world. If you want to be part of that process, reach out to us. There is a room in our team for exceptional researchers — and not only computer scientists, physicists or mathematicians. To solve interdisciplinary problems, we need an interdisciplinary team. Among others, we seek economists and social scientists familiar with game theory, systems governance and stability, redistribution, and mechanism design. If you have expertise in those or related fields, don’t hesitate to apply.


¹ A consensus protocol is a means of achieving an agreement in a decentralized systems, where participant’s honesty can not be assumed. It is one of the fundamental problems for distributed computing and multi-agent systems.


References and links:

[Abadi 2018] J. Abadi and M. Brunnermeier, Blockchain Economics, Working Paper, pp. 1–67, 2018.

[Anderson 1972] P. W. Anderson, “More is different,” Science, vol. 177, no. 4047, pp. 393–396, 1972.

[Budish 2018] E. Budish, The Economic Limits of Bitcoin and the Blockchain Bitcoin and the Blockchain: A Critique in 3 Equations, NBER Working Paper №24717, 2018.

[Spałek 2013] J. Spałek, Emergence in Laws of Nature and Hierarchical Structure of Science, chapter in the book How Science Spies on Nature and How Technology Imitates Nature; edited by A. M. Kłonkowski & M. Jaskuła; Publisher: Wydawnictwo Uniwesytetu Gdańskiego, Gdańsk 2013, pp. 147–159.