Long time no posting! I’d like to write about two updates since my last post here in 2018: my Master’s thesis at fortiss and my work as blockchain developer for ETH Zürich.
Ontologies
My thesis is meanwhile published here. It was a great foray into the world of ontologies and semantic reasoning. One that will certainly inform my coding and thinking from now on, maybe also my career choices. In short, it was about matching the capabilities of a robotic gripper with the geometric conditions of an object.
The gripper in this screenshot for instance has a few options how to grab different objects. In classical motion planning, millions of possible grasping approaches would be computed to find the few feasible ones. The interesting part about leveraging geometrical data to a semantic level, is, that it can be reasoned with. With a few queries the software can “reason” about grasping options. Just like we would intuitively know how to grasp objects with our hand. What I like about reasoning with ontologies as a way to comb through data in contrast to (most) machine learning approaches is, that backtracking is always possible. It makes rarely sense to “ask” a neural net how a certain outcome came to be. A semantic reasoning system however, always offers perfect insights into its workings. A memorable example that my thesis advisor gave me in the beginning was, that once we are able to reason about what a gripper can do with an object, we could just as well reverse the queries and ask how a new gripper might look like that is optimised for grasping a set of certain objects. From my experience of expanding the ontology of fortiss with these capabilities, I gained a huge respect of the power these approaches can hold. The laborious process of designing ontologies pays back abundantly through the many obvious as well as unforeseen ways it can bring added value.
Blockchain
My curiosity about the blockchain space was building up via the ZigZag podcast, Munich Legal Tech events and other sources. When I saw a seminar at TU Munich about it, I signed up despite not getting credit points for it in my Masters. In this seminar, among others, Dr. Marcus Dapp from the chair of Computational Social Science at ETH Zurich offered a project related to his research project “Finance 4.0“. After our team assignment was done in the summer semester 2019, I continued working on it under ETH-employment: four months living in Zurich and then from the beginning of 2020 to the (preliminary) end in July remotely from Munich. It was a very steep learning curve and I enjoyed it a lot. Getting into a new technology from scratch more or less by my own helped building confidence. 
The basic idea of Fin4 is to foster positive sustainable actions, be it in ecological or social space. Anyone can create “currencies”, called tokens, using our decentralized app (dApp). I could create a “Clean the Isar token” that can be claimed by providing photo-proof of a cleanup action. If people are interested in claiming this token depends what kind of ecosystem emerges around it. The city of Munich could decide to waive some amount of taxes in exchange for sustainability tokens claimed by citizens. And maybe I can trade my Isar tokens with some Spree tokens when I move to Berlin. The idea is, that by enabling currencies to spawn on any level of granularity instead of just one big Euro or one big Dollar, a multitude of “economic currents” (see the similarity to “currency”) can be incentivized that were previously hardly visible or entirely inaccessible. Local actors anywhere can start building ecosystems of tokens around needs they see in their communities for fostering sustainability or social cohesion. Among the first questions people have about the project is: why blockchain? While I wouldn’t answer this question in the affirmative for a lot of blockchain projects out there, I would definitely do so for the Fin4 project. I would say it’s a perfect ideological match. Let’s say a major in a small village in Italy wants to incentivize the farmers in his village to use less water on their fields. The project has a timespan of 10 years. With a centralized Fin4 architecture he would have to trust the ETH Zurich to keep the servers and the database running for that time and not to tamper with the data. But with the decentralized architecture we chose by making it a dApp on the blockchain (an Ethereum test network for now), he has to trust no one to keep things running. Once deployed, the smart contracts (code on the blockchain) can never be removed or tampered with. All code and all transactions are visible for everyone forever. There would be so much more to say about the project – if interested, check out the project website finfour.net that also links to an extensive eBook and a live demo that I worked on. I am grateful to have had this opportunity to deep-dive into the blockchain space with such a profoundly interesting project and feel enriched with ideas and lots of new useful skills.
Benjamin Aaron 