“They're writing an essay about institutional memory.. Something that used to take 30 minutes took 30 seconds. They looked at the old way and couldn't believe they'd tolerated it. That was the aha.”
When I'm writing an essay about institutional memory, I want to capture ideas in a way that connects them to everything related, automatically, so I can build a knowledge base that gets more useful the longer they use it — compounding, not filing.
A researcher, academic, writer, or knowledge-intensive professional who uses Roam because it is the only tool that treats the connection between ideas as a first-class object. They write in Daily Notes. They [[bracket]] everything. They have a graph with 3,000–15,000 nodes that they've been building for 2–4 years. They know their graph is their most valuable intellectual asset. They also know that Roam's development has slowed, that the tool has rough edges, and that they've considered migrating to Obsidian or Logseq at least twice. They haven't migrated. The switching cost is partly the data — mostly the habit.
To reach the point where capture ideas in a way that connects them to everything related, automatically happens through roam as a matter of routine — not heroic effort. Their deeper aim: build a knowledge base that gets more useful the longer they use it — compounding, not filing.
roam becomes invisible infrastructure. Capture ideas in a way that connects them to everything related, automatically works without intervention. The old problem — roam's development pace — features requested for years, still not shipped — is a memory, not a daily fight. Performance optimization for graphs above 5,000 nodes restores the fluency.
They're writing an essay about institutional memory. They type `[[institutional memory]]` and see the linked references panel populate: 23 notes they've written over 3 years that touch this idea. A note from 2022 about how hospitals manage handoffs between shifts. A note from a book about Inca quipu. A note from a conversation. Three of these connections surprise them. One of them becomes the essay's argument. They didn't plan any of this. The graph did it. This is why they haven't migrated.
Has been in Roam for 2–5 years. Writes Daily Notes consistently. Has 3,000–15,000 nodes — pages, block references, linked mentions. Uses Roam for research, writing, meeting notes, and evergreen note development. Has a consistent naming convention for certain page types: authors, books, concepts, people. Uses `{{query}}` blocks for dynamic views into the graph. Has recommended Roam to colleagues; few have adopted it. Has a backup system (Roam's export, Dropbox sync). Reads the Roam community Discord occasionally. Has filed a feature request. Still waiting.
The proof is behavioral: capture ideas in a way that connects them to everything related, automatically happens without reminders. They've customized roam beyond the defaults — templates, views, integrations — and their usage is deepening, not plateauing. When new team members join, they hand them their setup as the starting point.
Not a feature gap — a trust failure. Roam's development pace — features requested for years, still not shipped happens at the worst possible moment, and roam offers no path to resolution. They open a competitor's signup page not out of curiosity, but necessity. Their belief — most note-taking is filing, not thinking — Roam enforces the difference — has been violated one too many times.
Pairs with `obsidian-primary-user` for the Roam-hosted vs. local-file linked knowledge graph comparison. Contrast with `logseq-primary-user` for the proprietary-hosted vs. open-source-local networked thought tool philosophy. Use with `readwise-primary-user` for the Readwise-to-Roam highlight pipeline that feeds the graph with external reading.