Hydra is just one potential puzzle piece for me. What’s driving me is the idea that there might be a way for computer-based systems to simply not be annoying. With “annoying” I mean the flood of different login methods, the heavily ununified UIs, incompatible database schemas, constantly needing to switch between so many different windows and tabs , buggy software, needing to be connected to the internet constantly and everything else that causes friction when needing to interact with digital systems.
This vision that’s constantly making be bounce between excitement, insanity and hopelessness is that there might be a formulation that brings order into the chaos, not by imposing a certain structure or language but by providing a way for all our existing structures and ways how we put the world into symbols to become translatable into each other, to lay the groundwork for equal global cooperation across borders. The only thing I properly studied was computer science, so I’m really a complete noob when it comes to category theory, but I have a feeling that somewhere inside it, there’s a solution to my problems. I found https://topos.institute/work/catcolab/ yesterday too, and even just “technologies that enable cooperation across difference” from their homepage is basically exactly what I’m also all about. I want to find a way to make software engineering in general “become category theory” (sorry I cannot express it better yet).
I believe that the next big paradigm shift in computers will be leaving Unix’s “everything is a file” behind and transitioning to “everything is a graph”. I think everyone should have at least one personal graph stored with a private key that’s exclusively under their control, and then enabling event-driven/streaming/real-time selective disclosure between different graphs for secure data sharing. Since the beginning of this year I’ve gotten deeply into RDF, but of course that’s also just one of the puzzle pieces.
On the data storage side I’m therefore looking for formulations of everything that makes up a computer program in terms of graphs. The data side is rather trivial. For the business logic side, what Josh is doing with LambdaGraph and Hydra might be suitable. On the UI side, I discovered Goguen’s “algebraic semiotics” yesterday, but didn’t have time yet to get deeper into it. However how to formulate the runtime also in terms of graphs and possibly category theory is still a mystery to me.
Idealistically, what I care about is legibility, particular to make governments, companies and other organizations truly legible and understandable, to get away from our contemporary messy black-box institutions to making the processes that affect our lives auditable by everyone. Laws should not require “today’s priests” (i.e. tax advisors, lawyers etc.) to be understandable and actionable, but instead be clearly divided into what’s mechanical/deterministic and what requires human judgement, so that the execution of governance becomes a creative, compassionate, social dance again instead of so many people serving the system like mindless computers. I want to put an end to “bureaucracy anxiety”, which I call the pathological state of settling with exploitative working conditions simply because the fear of doing something wrong during the bureaucratic processes necessary for self-employment or founding a company are too overwhelming.
On the collaborative side, my vision is something similar to the existing “Wikifunctions”, so as the next generation of open source I see that we’ll be transitioning away from entire git-based projects (because they always reinvent the wheel regarding data storage, formulation of business logic and UI) and rather have a collaboratively edited catalogue of functions. Or maybe it would be a catalogue of effects, or of morphisms (I’m not well-versed enough in functional programming yet), but somehow it would be a sort of “Wikipedia of what computers can do”. I had already thought about “what if we treat data as carefully as real goods?”. For that I recently got more deeply into GS1 EPCIS, that’s the data storage and exchange standard that coordinates what eventually ends up in your local supermarket and practically all other global shopping and logistics, as well as Industry 4.0 digital twin environments like Asset Administration Shell and Eclipse Basyx, in an attempt to see data ingestion, transformation and output as equivalent to industrial production.
Notably liberating data from legacy institutions will also require a great deal of reverse engineering of web APIs, not just binaries like “Reverse Engineer Anything”. With Divizend we already reverse-engineered the web interfaces of dozens of banks and tax administrations to allow users to seamlessly import their stock portfolios and documents into our platform to pre-fill their reclaim forms and later submit them with a single click, so while we’re still in our contemporary world with all the legacy systems which often don’t have proper APIs, reverse engineering is the only option.
Practically, where I want to start with is tax filings. Through various coincidences in my life, I became co-founder and CTO of a company which sells a very specific kind of enterprise tax filing and regulatory tax compliance software (focused on withholding tax reclaims for foreign dividends and interest) not only to tens of thousands of private individuals but also many large international banks and tax advisors. I therefore have complete creative freedom what I want to bring into the market, also because we’re completely self-funded and have no external investors who would impose any other strategy onto us. Right now our platform is based on traditional 2020s-era web development technologies like Next.js, MongoDB, Sentry, etc., with all the different features bolted on over time, but right now I’m working on the architecture for this radical graph-based, category theory-based re-engineering. Regulatory reporting is where all the features I want to make accessible to private people are already mandatory: strong encryption, auditability, traceability, secure data sharing between organizations, on-premise if desired etc. Here I want to get to a point where I can formulate general bureaucracy and thus also regulatory reporting frameworks in terms of graphs and to somehow use category theory to not lose the overview when handling this massive complexity.
If you have any pointers for me that might be relevant to this research question of "How can we make the structures through which humans coordinate their lives explicit, understandable, interoperable, and changeable, without requiring everyone to agree on one universal representation of reality?", please let me know too :)