Show HN: Sutra-OS – A self-hosted ontology platform
github.comI beg the authors to reevaluate their stated belief in a system that promotes data, logic, and actions as fully representative of decisions. Most human-made decisions are driven by feelings and then our forebrains spin narratives about the logic involved.
The most powerful technology we can deploy is the capability to identify, process, and communicate our feelings.
I am intrigued but want to know more about how I might use it. Suggestion - lots of people like me are heavy airtable/coda-like users who setup relational DBs for various projects. Should I be looking at this to replace any of these efforts?
it is not so much about replacing stuff (although also possible) but more about abstracting everything into a general layer above all your databases. you can then query and writeback this layer and sutra then communicates with all your databases. you talk human language (objects, relations, actions) and sutra then translates this into every dialect the computer needs (csv, sqlite, pg, mssql, mysql, your mails... extendable)
It’s turtles all the way down.
Compared to other turtles my turtles know each other and can act. Object-types, link-types, action-types - each having their instances. Batter my heart, three-person'd God...
Many small ontologies, each held by those it describes, keep power spread out; one model of everything would not.
Same with self-hosted vs SaaS or any other organizational structures that relate to people and power over them.I feel bad for people applying ontology to their life and work. it seldom works, and fundamentally questionable.
(Agree that this seems like rage-bait -- ontology is literally just the process of deciding which abstract structures are useful for reasoning about a domain.
If you are working in a domain of knowledge, how can you possibly operate without ontology?)
If you use an object-oriented programming language and create a class, you're doing ontology.
If you are using a relational database and you create a schema, you're doing ontology.
If you choose to use an object-oriented programming language or a functional programming language, you're doing ontology.
This has got to be ragebait. If you do any serious software engineering, ontology is inescapable. Eg. I would like you to model software for the automotive industry without using ontology. Try it. There's a ton of concepts, terms, hierarchies to get through before you write a single line of code. Without ontology, you'd be shooting yourself in your foot every 5 minutes, I guarantee you that.
Completely agree, ontology is inescapable if you have a problem to solve, an application to model, but totally useless if you only want to model Ding an sich.
It's a trace of trajectory, not the thing-in-itself.
I understand not applying it to life, but have you run into specific issues with ontologies at work? Structured categorization and organization of information is good on first principles to me, provided that you can maintain it.
> Structured categorization and organization of information is good on first principles
No, it's probably wrong. I don't think anyone able to model(structure) `user` properly without the context of the application.
Hm. I think there are some commonalities that you can model within a company that guide how you should structure information for that company, but I think we might be talking about applying an ontology in different ways. If you don't order the information at all, it becomes impossible to retrieve when you need it.
It is wrong in the same sense that Hilbert's program was wrong, but mathematics still thrives as well as it's effectiveness.
but it might work now with AI?
that was because they had no idea what they were talking about :)
Ironic, given the software world's misuse of the term ontology
Nice to see an implementation that includes Action Types. Most implementations miss this and only deal with Object and Link Types
I really appreciate your comment because this is a key part for understanding the idea.
Can't you use obsidian to build ontology through markdown and interlinks?
the point is that sutra-os has also typed actions. that means you don't only link stuff but you execute the linked stuff against each other. it is like human language executed like lisp.
How does this compare with something like Datalog?
Few people think and reason with rigorous ontology in real life. That's why ontology is hard to popularize—but it's also exactly why most people drift through life mediocrity intact.
How does ontology help people be exceptional?
Written in clojure and also looks like Farmlogs - wild. Do I know you? lol
oh I researched what farmlogs is - no, that is a coincidence ^^ could have used any other example too like aviation or stuff
Huh. I’ve never read a code project page on GitHub that sounded like a Keynote speech at a developer conference. Not sure what to think about that as a reflection of the project’s design, use and governance. Other than I’m supposed to conquer the world using this software?
I’m not familiar with ontology software. So I will continue to read this and see if I can learn how I might grow my handwritten taxonomy into something actionable for my wiki and prompt repo.
If you only want "grow my handwritten taxonomy into something actionable for my wiki and prompt repo" than Sutra-OS is not the right tool. I would use Logseq for such a thing. If you feel the limits of stuff like Logseq in the future you should explore Sutra-OS. It extends the idea of a knowledge-graph to actionable-knowledge-graph.
> Ontology-Oriented Software Development, whose creators I deeply admire
Links to Palantir. Ontology means legibility for the empire, yet again. A sad laugh escapes my other mouth, and I close the link.
Readme is covered in slop hallmarks. LLM prose, LLM UI design. This just means I can generate something better and more preferential myself. The experience doesn't reflect any understanding of the domain beyond what I could generate. Not interested. Sorry.
pull requests are very welcome to make it less slop :)
The maintainer sets the tone. Contributor pull requests, if merged, may temporarily improve the state of the repo, sure, but they're much less likely to influence the maintainer's quality bar for the repo.
And so the current state of the repo, before anyone else contributes, must communicate that tone, because that is a trust-building exercise. What you have to communicate is, "I will cherish your efforts; I will not let them go to waste." The AI-isms in the repo currently communicate the opposite.
If you use a building you have respect for the architect, room-decorators, facility managers, cleaners and so forth. Doesn't mean a cleaner will expect room-decoration from the architect. Also almost nobody cherishes effort - only the work that gets done when playing a certain role for a certain time.
Well, if you want contributors, maybe just run a premortem on the goal itself. That could sound like this: "It's a year later, and this project still does not have contributors. What went wrong?" Then try to exhaustively list what could have gone wrong, rank those things by expected probability, and ask yourself what you can do now to make each of those things go right. Good luck.
As per the HN idea, I shared my repo here as a gift and for others to examine something I find “deeply interesting.” I didn’t share it with a sense of communal entitlement, wishes, assumptions or expectations from me toward others or vice versa. Self-congratulatory, imaginative (like the first comment in this thread "beyond what I could generate" - show me your code) or lecturing comments are not “deeply interesting” in that sense.
you don't have to do it by hand
try https://github.com/blader/humanizer or https://github.com/etodanik/deslop