Hello! My name is Jeremy, and I’m glad to be here and get more involved. A bit about me:
I’m a maker! My undergraduate studies at UMass Amherst focused on creative mechatronics (robots, software, circuits, mechanical design) and I worked heavily on development of two academic makerspaces. I recently finished my Master in Design for Emergent Futures at Fab Lab Barcelona/IAAC/Elisava.
I am interested in developing the sociotechnological infrastructure necessary to provide equitable and agile alternatives to the existing paradigm, especially in education, fabrication, agriculture, and, most notably, information (more on that later). I am deeply concerned about the trajectory we are on as a species, and it is clear to me that with the seemingly exponential development curve of technology, the future will not be like the past. I think we need to start preparing proactively now to get a more agile alternative social/governance infrastrucure in place to mitigate the damage before the pace of technological change becomes significantly more than our existing institutions can handle.
I think that strengthening the resilience of local communities by moving toward self-sufficiency and circularity will be part of the solution. I am motivated by the possibility for communities to leverage new technologies to provide for their needs and produce what they consume while supporting and being supported by peers in a globally cooperative information network.
For this to happen, we need better software infrastructure for working with and sharing information. We need accessible information and workflows to enable people to launch community initiatives; develop, govern, and run local makerspace organizations; and share, collaborate, and publish insightful results. Individuals need to feel empowered by the tools and information available, not overwhelmed. The Internet was supposed to democratize information, but today our information is owned primarily by a few app silos that compete for our attention. Many struggle with the barrage of information, and we lack sufficient tooling and workflows to process, build, and retrievably store what is valuable in an interoperable way. I’ve concluded that existing tools just aren’t good enough for the interconnected work that needs to be done in the near future.
My software development project, Seption, aims to provide for these needs by taking a new approach to information management in computers. Well, the concepts individually are actually quite old, but nobody to my knowledge has combined them like I’m planning. I envision it as more of an open protocol than a platform.
The plan is to have a decentralized and content-distributed semantic graph database that makes heavy use of transclusion to let users model and visualize nonlinear thoughts, compose documents, store and link data, collaborate with others, make decisions, track provenance, and publish to the Internet, while being open source, interoperable, permissioned, secure, and highly extensible. Seption should be radically compatible with existing tools, file formats, and integrations.
My approach for the core architecture is to have a simple set of rules from which the features of the software can emerge, which leads to flexibility, expandability, and reduced software development on hardcoded special cases.
Seption will let users create meaningful connections between all the information they store on their computers (messages, tasks, calendars, contacts, documents, images, audio, video, and anything else stored natively, in files, or with a URL), so that things can be easily found via other conceptually related things.
Everything has to do with everything else. All information you store in your computer has something to do with other things you also store there. The complaint letter you write to your online bookstore is a response to the overly high bill they’ve sent you. The photo your best friend emailed you was taken on your shared holidays in Italy, and the poem you’ve written yesterday was also conceived there. The Aunt Mary whose birthday you enter into your calendar is same as the Aunt Mary in your address book, the same person that borrowed your book about old English sheepdogs (which you have to remember to ask back from her). —Vision of Fenfire
In the past few decades we’ve seen exponential growth in data and information. However, our ability to make sense of these inputs, to convert data and information to knowledge and wisdom, is already capped by the mainstream tools today.
Using a graph database, bi-directional links, and hypertext transclusions, these tools enable users to create and traverse knowledge in a way that mirrors how humans organically create knowledge — associatively and contextually.
These “knowledge graphs” break out of the “file-and-cabinet” hierarchical paradigm that most computer systems use — note-taking apps of the last few decades, filesystems, HTML documents, etc. Users using Tools for Networked Thought can more easily create meaningful relationships about the world we live in. Not only that, these users can then more easily recontextualize relationships, allowing ideas to compose and refactor in emergent ways. This leads to the creation of more insights, more interdisciplinary insights, and generative insights: insights that create more interdisciplinary insights. —Vision of Athens
I am building on the shoulders of giants: Vannevar Bush’s writing on the Memex, Ted Nelson’s Project Xanadu (and ZigZag), Sir Tim Berners-Lee’s Semantic Web, and concepts championed by dozens of contemporary software projects, including Amplenote, Appflowy, Arena, Athens, Capacities, Curio, Docear, Fenfire, Glue, Gruff, Heptabase, IA Writer, Intrigue, Kanopi, Kosmik, MarkMind, Milanote, Notion, Obsidian, Protégé, Qiqqa, Readwise, Roam, Scapple, Scrintal, Scrivener, Semilattice, Solid, Speare, Subconscious, Tana, Tangent, The Brain, Todoist, and Webmemex.
The most similar projects to Seption linked above have failed. I think I have a better chance at success because the scope is limited to developing the core architecture, and any potentially enticing scope creep will be relegated to implementation via plugins after the core has been implemented. Also, I could be naive, but I believe that it hasn’t really been possible to do this until now. With Large Language Models, word/vector embedding technology, many relevant recent W3C standard recommendations, open source decentralized syncing protocols, the superior compute power of today’s devices, and the option to boost performance further for some types of operations with new web technologies like WebGPU and WebAssembly, I’m confident that it can be done.
In a soon-to-be world with personal AI agents, I think it is more important than ever to have personal semantic graph databases- to be able to label and track AI generated content, as well as to give AI agents a way to store information that is useful, familiar, and navigable by humans and other AI agents.
This is the first public post about this project, and things are still fairly early. At this point, Seption is toward the middle of the conceptual phase. I’ve been doing research on prior art and working on design documents that go into technical detail about how the architecture, data structures, and algorithms will be implemented to provide the features. I plan to pretty these documents up and build a website (I’ve acquired seption.org), and then start development on a minimum viable product, which will eventually round out into the “core” leaving most everything else to be implemented as plugins, as mentioned previously. I’d also like to secure some funding and start a nonprofit.
I welcome thoughts, feedback, advice, and recommendations. I have started a Discord server for discussion about the project. If it is appropriate to start a thread on this forum, I am open to that as well.
