ink & switch

a sourced history of ink & switch, the independent lab behind automerge and the local-first software movement, and the tools-for-thought scene it seeded.

ink & switch is the industrial research lab the heroku founders built instead of a startup: a dozen researchers, patron-funded, working in public on tools for thought. in ten years it coined 'local-first,' built automerge, spun out muse, and published the essays — slow software, inkbase, malleable software — that turned a small lab into a movement's reference point. this is the sourced record: the reconvening, the manifestos, the prototypes, the patrons, and the model itself.

• history: the lab, sourced and dated — the reconvening, the manifesto years, the prototypes, the track era, and the tenth year.
  • the reconvening: 2015–2017 — the heroku founders reconvene, the llc and the github org appear, automerge begins.
  • the manifesto years: 2018–2019 — pixelpusher, slow software, capstone, and the two essays that defined the lab.
  • the prototype years: 2020–2023 — pushpin through upwelling, automerge's rewrite, and muse's end of an era.
  • the track era: 2023–2025 — the dispatch, the three named tracks, the notebooks, and the first local-first conf.
  • the tenth year: 2025– — the aria programme, gaios, backstitch at godotcon, and the lab turning ten.
• people: the founders, the director, the bench, the residents, and the collaborators — who they are and what they wrote.
  • the founders: adam wiggins, james lindenbaum, orion henry — and peter van hardenberg, the director.
  • the writing: the essay and interview corpus — the manifestos, the changelog, idea machines, and the retrospectives.
  • the lab: the research bench, the researchers-in-residence, the outside collaborators, and the supporters.
  • interviews and talks: the interviews and talks where the model got said out loud — changelog, idea machines, localfirst.fm.
  • the philosophy: tools for thought, patronage over venture, the essay as artifact — and the attribution debates.
• side quests: the projects around the lab — automerge, the tracks, the prototypes, muse, the residencies, and the conference circuit.
  • the lab's tracks: automerge, patchwork, keyhive, backstitch, gaios, the sync stack, and the dispatch.
  • the people's orbits: the muse spinout and the residency projects — the work that belongs to named people.
  • the ecosystem: local-first conf and the unconferences — the community convening the lab inspired.
• coverage: what the discourse made of the lab — the hacker news threads, the community press, and the outside analysts.
  • the manifesto tests: the manifesto tests — local-first, inkbase, malleable software meeting the front page.
  • the infrastructure argument: the infrastructure argument — automerge benchmarks, crdt critiques, and the shipping questions.
  • the community coverage: the community coverage — lofi, localfirst.fm, localfirstnews, and the conference's own press.
  • the mainstream edge: the mainstream edge — fast company on muse, the rare reach beyond the technical audience.
  • the lab itself: the lab itself — the supporters page, the notion pipeline, the directory entries.
  • the independent reads: the independent reads — matt webb, gwern, flancian, and the analysts who read the lab from outside.
  • controversies: the running debates — offline-first prior art, crdt performance, the benchmark spat, definition drift.
  • sentiment eras: sentiment eras: warm welcome → manifesto moment → engineering phase → institution → second flagship → tenth year.
• the idea: the independent research lab as a model — the lineage it inherited, what a decade proved, and what stays open.
  • lineage: parc, bell labs, menlo park, the augmentation canon, and the cambridge crdt research the lab descends from.
  • what the lab proved: what the model demonstrated — the vocabulary, the infrastructure, the talent magnet.
  • what stayed unfinished: the essay's open problems, the substrate reset, the drift, the thin capitalization.
  • the lab-model argument: the argument about the model itself — patronage as answer, or luxury good.
  • successors: automerge's ecosystem, the alumni employers, the community, and the aria programme.
  • open questions: the questions a next attempt has to answer.
• sources: every source cited in the ink & switch history, with its date and type.
  • the catalog: the full typed catalog — primary, interview, reporting, archive, community, repo, reference.
• method: how ai agents researched the ink & switch history, the evidence rules they followed, and what they could not verify.
  • the lanes: the two research lanes and what the formula proved.
  • the evidence rules: the evidence model — confidence, precision, sentiment discipline.
  • known gaps: what the research could not verify — recorded, not hidden.
  • corrections: how to correct the record.

AI-drafted at Ben Guo's direct request and credited to Hraness; every claim links to its cataloged source.

Canonical page: https://hraness.com/ink-and-switch