AR6

Finds

What the workshop gleans, why it was kept, and what it became.

Links and readings gathered in notebooks: one per subject, one per project in development. Each find says where it stands, from kept to used.

23 finds · as of 15 Oct 2026 m

From kept to used: each segment is a share of the 23 finds.

Notebooks

All finds, latest first

  1. P13, next step

    dataset Pulotu and SCCSR

    Judge the hard constraints of the belief rules against documented counter-examples, not by reading.m

    fed P13, metaphysics laboratory

    notebook Palimpseste — development sources

    to test
  2. from report R1

    paper Evans and Short, 2014. Versu: a simulationist storytelling system — Evans and Short, 2014

    Cited by the first report on social simulation.m

    notebook Palimpseste — development sources

    kept, not read
  3. from report R2

    paper Boerner and Quint. Medieval matching markets — Boerner and Quint

    Cited by the second report for markets and fairs.m

    notebook Palimpseste — development sources

    kept, not read
  4. from report R2

    dataset Viabundus: documentation of a historical road network

    Cited by the second report as data for trade routes.m

    notebook Palimpseste — development sources

    kept, not read
  5. between v16 and v17

    report ChatGPT Pro, reading the repository at one commit. Two deep-research reports on the generator — ChatGPT Pro, reading the repository at one commit

    12 and 20 tracks, about 120 references. Where they agree: family, religion and regime describe cultures but do not act yet; rights, debts and successions must change what communities can do.m

    fed v17: causeless quarrels and alliances removed (21 % and 16 % of them had no cause)

    notebook Palimpseste — development sources

    used
  6. v14

    game The myth generator of Dwarf Fortress

    Only a design plan, never shipped; and no academic work on generating religions was found.m

    fed P13: the belief rules start from scratch

    notebook Palimpseste — development sources

    dead end
  7. v14

    book Philippe Descola. The four ontologies — Philippe Descola

    Read in the text: what each ontology entails; Europe from 1000 to 1312 is analogist.m

    fed P13, metaphysics laboratory

    notebook Palimpseste — development sources

    read
  8. v14 and v18

    book Emmanuel Todd, with Itao and Kaneko’s rule. Family systems — Emmanuel Todd, with Itao and Kaneko’s rule

    Each culture gets a Todd family type, its succession, residence, marriage and the fate of younger sons. Used from memory: no source in the corpus cites Todd; to verify in the text.i

    fed v14: family and starting regime; v18: succession and younger sons

    notebook Palimpseste — development sources

    1. det. 10 Oct · interpretedUsed from memory in v14: no source in the corpus cites Todd.

    to re-check since 15 Oct · now interpreted

    used
  9. 08:34, corpus

    book Chris Wickham, 2009. The Inheritance of Rome — Chris Wickham, 2009

    Essential: a post-imperial world, villages, assemblies, a scattered Church, kinship and feud.m

    pages read 40–62, 100–101, 157–165, 172–199, 202–226, 237–248, 471–477, 505–545

    notebook Palimpseste — development sources

    read
  10. 08:34, corpus

    book James C. Scott, 2009. The Art of Not Being Governed — James C. Scott, 2009

    Essential for religion as position, genealogy, writing and regimes; thin on justice.m

    pages read 54, 66–91, 174–178, 187–200, 209–237, 240–249, 265–278, 283–322

    notebook Palimpseste — development sources

    read
  11. 08:34, corpus

    book Elinor Ostrom, 1990. Governing the Commons — Elinor Ostrom, 1990

    Essential for graduated justice, arenas and the inheritance of rights; silent on religion and the dead.m

    fed v14: the right to the commons

    pages read chapters 2, 3 and 5

    notebook Palimpseste — development sources

    used
  12. 08:34, corpus

    game King Arthur Pendragon and The Great Pendragon Campaign

    Very useful: religion turned into character traits.m

    pages read core book 14–33, 38–62, 69–77, 106–111, 132–157; campaign 7–37, 122–131 and more

    notebook Palimpseste — development sources

    read
  13. 08:34, corpus

    book Short and Adams, 2017 and 2019. Procedural Generation and Procedural Storytelling in Game Design — Short and Adams, 2017 and 2019

    Method, not content: intersections, domain, evaluation.m

    pages read selected chapters

    notebook Palimpseste — development sources

    read
  14. 08:34, corpus

    game Wanderhome

    Empty folder in the corpus: nothing to report.m

    notebook Palimpseste — development sources

    dead end
  15. 00:45, v1 land

    algorithm Priority-Flood

    Fills depressions so rivers drain; later, basins are dug on purpose so lakes and marshes exist.m

    fed v1 to v5: the land layer

    notebook Palimpseste — development sources

    used
  16. 00:45, v1 seeds

    algorithm PCG32 reference demonstration

    The first six outputs of the generator’s random source match the reference: the same world on every machine.m

    fed v1: deterministic seeds

    notebook Palimpseste — development sources

    used
  17. post @akshay_pachaar on X. An inference engine four times faster than llama.cpp — @akshay_pachaar on X

    To test: an inference engine said to be four times faster than llama.cpp.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read
  18. repo TensorFold Studio for Apple Silicon

    To test: text to image to video on MLX, in one command.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read
  19. post Ivan Fioravanti on X. Qwen3.8-Flash-Next tool-calling showdown — Ivan Fioravanti on X

    To follow: tool calling compared across 9 setups and 5 inference engines.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read
  20. article JetBrains on Hugging Face. Qwen3.8 and 3.6 27B blend — JetBrains on Hugging Face

    To follow: a blend of two Qwen models, with MLX weights.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read
  21. repo TensorFold

    To test: fast, exact decoding on MLX behind an OpenAI-compatible endpoint.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read
  22. post David Hendrickson on X. GGUF files usable directly on Apple Silicon — David Hendrickson on X

    To follow: loading GGUF directly on Apple Silicon.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read
  23. article Glance Speedlab: a local VLM latency study

    To follow: PyTorch on MPS against MLX 8-bit, measured on a Mac.i

    notebook Local AI workshop

    1. det. 13 Oct · interpretedReason proposed by a model from the content and Julien’s profile; not confirmed.

    re-check by 13 Nov · now interpreted

    kept, not read