like: (I, read, Q8337) → generates triplet ID: T12345 (Friend, likes, T12345) (Friend, comments, T12345) (I, responds-to, Friend-comment) it elegantly solves the "threading" problem that plagues most social media. conversations naturally branch and nest without requiring special design. this is essentially how modern graph databases handle complex relationships, but applied to social interactions. it's remarkably elegant because it matches how humans naturally think about connections - everything can reference everything else. and the beauty is how lightweight it is. no need for complex schemas or rigid structures. new verbs/predicates can emerge organically as needed. want to track "abandoned halfway through" for books? just add it! the system adjusts.

this approach keeps the model clean while allowing unlimited richness. your thoughts become first-class objects that can themselves be referenced, commented on, or connected to other concepts.

imagine being able to analyze not just "how many people read dune?" but "how do readers' emotional responses to dune evolve over time?" or "what patterns emerge in how people connect seemingly unrelated works?"

the dream would be a system where tracking your experience feels less like "data entry" and more like "leaving traces naturally as you move through digital spaces."

it's essentially creating a map of human experience

imagine if every app implemented something like "activity streams" where your interactions were accessible in a standard format. suddenly connecting these experiences becomes possible without elaborate custom integrations.

i just realized that a software company (say twitter), is nothing but a UI and fast indexing over a particulr “section” of this data. all tweets by a user is nothing but all triplets such that (q== user, action=posteed, q.has_property(is of type, tweet)) for example. software companies are nothing but efficient providers of this

build with centralized infrastructure for immediate performance needs, but design the system to be blockchain/decentralization-compatible from day one.

what's common across these examples is that the founders were all solving a problem they personally experienced deeply, started with a minimal viable product, and had a clear vision of how things could work differently. the initial implementations were often quite basic, but the underlying model was sound - which is exactly where you are now.

I'm building a universal graph for personal experience - think of it as a protocol for how humans interact with information. It's a simple triplet model (I, read, book) that can capture literally any interaction or thought you have, while connecting to canonical entities like wikidata. Unlike siloed apps where your book data is separate from your music data, this creates a unified knowledge graph where everything connects naturally. The beauty is that software can then just be views over different slices of this graph - reading apps show your book nodes, music apps show your listening nodes, but the underlying data structure is consistent. It starts with personal tracking but evolves into a social layer where friends can reference your experiences or build upon them. The system treats your thoughts and activities as first-class objects that can be referenced, connected, and discovered over time. We're solving the fundamental problem of digital fragmentation by modeling data the way your brain actually works - as a network of connected experiences rather than isolated apps and feeds. Think of it as the infrastructure layer for your digital life, with apps being just different lenses into your personal knowledge graph

you don't just want a better book tracker or opinion logger - you want a unified system that handles it all because that's how your mind naturally works.

properties for this type

properties for this type

what you're describing is essentially a bottom-up approach to ontology building that mirrors how language itself evolves

in practice, this might look like: each user implicitly creates their own ontology through usage the system notices when different users are tracking similar concepts common patterns get standardized and suggested to other users users can always extend with personal concepts that matter just to them

Constraints

yes, wikidata's underlying software (wikibase) is open source! you've got two main routes: for a personal fork: you can install your own wikibase instance (what powers wikidata) this lets you use the same data model but with your own rules about what's notable/acceptable the code is on github, maintained by wikimedia foundation many organizations already run their own wikibase instances for specialized knowledge that doesn't fit wikidata's notability guidelines.