Skip to main content

Course Notes ('19)

Published onJul 08, 2020
Course Notes ('19)

Collaborative notes from Structuring Collective Knowledge, Ep.1 / slides


Background examples:


Knowledge Futures Group:
watch for updates on instagram: @knowledgefutures

SJ (KFG, Wikipedia)
Jabari (U: knowledge graphs + systems),
Agnes (G: organization and distributed forms of knowledge),
Sarah (G: computational neuro; models for structuring concepts in brain + mind, and ways to generate art),
Jack (UL interest re: attaching prov to data, RWoT, and sensor attachment),
Robert Solomon (early ML, multidim thinking, viteracy [js programmers wanted!], enhance potential of the brain. making explicit what’s only implicit),
Justin (Machine Intelligence Community).
MM (Retired chemist): sharing lab bench data
Ed Bice (Meedan, Bridge (translation), electionland + Check: fact-checking, standards)
Dan Li (Arch + computation: generating 3d from photos)
Beth (Computational Law + DAOstack)
Siranush Babakhanova (U: knowledge (graph) temporal visualization + alternative time/calendar visualization f/ex for lab work)

Lightning talks

+ : “one characteristic image per square km”. simple runaway success, but largely limited to the UK.
+ fMRI alignment : fitting data into a shared space across people. Talairach space (1980s), CVS space, fs-average space. No agreement in the field how to do this.
+ ‘your friend the slime mould’ : physarum and dictyostelium. (quite different)
polycephalum: many cells aggregate when food is scarce, retain nuclei but share plasmoid soup. solves coordination problem without central computer
+ Ed Bice, Meedan Check:
human model of coordination, requiring [awareness, respect, attribution];
collaborative online models should reflect this

Tools (collaboration+) – SJ
We too are examples of such coordination: from cells to muscles (though we don’t like to think of ourselves as having lots of multinucleated things inside us, but we do)
from organisms to networks: termite mounds + road networks around the planet

classes of models
-decentralized vs. central (OSM v. Google Maps)
-catalytic (NASA datasets vs. Google Sky
-Alignment vs. standardization ( //HTML)
-Private service vs. common good ( vs. Iceland, seed vault)
-Feedback loops (selection, resonance)



  1. Model
    1:Describe what is possible (a representative photo of every km2!)

    2:Demonstrate a key example (my home town)
    2b: Name counterexamples (copyrighted aerial photos)

    3:Define how to share results (a map + webform + stats)

    4:Communicate, organize, iterate (a UK forum, outreach targets)

  1. Sharing Chemistry lab-bench data
    + 1: Coordinate and simplify sharing of methods
    + 2: Good examples — CRC handbook?
    + 3: Obstacles: Lab benches are too small; and they try to monetize “how to make a reaction go” or “how to make such and such a slurry” & sell it as a service. Patents: often limiting, hard to pile inventions on inventions.
    +4: Chemists against patents (look for advocacy groups)

  2. How do people use 3D digital model : collection of use cases [DanL]
    + The current 3D repos only have meta data for 3D geometry; (Check here to see the most common 3D repos.)

    + It will be great to see how people are actually using the digital model in their projects;

    + This project will be an extra layer of info/data that is super-imposed on the current digital model repo;

    + Project creators can get more attentions about their projects through these channels;

    + Needed: information architecture (how to structure the use case ‘knowledge’?) ;

    + Needed: tools. How to develop this Pinterest-alike website/app?

  3. fMRI alignment standards [Sarah]

    +comparing data and regions across individual brains requires alignment to a standardized space

    +need: standardized and reportable decision process for selecting template space, standards for sharing (+ realigning)

    +model: journal standards for publishing statistics (e.g. J Neurosci, academic orgs // APA)

  4. Large! exquisite corpse game. [Agnes+Jabari]
    + 2 ppl a minute are chosen (from a pool of people), and select one of 2 opts
    + Story forks, merge the result. Where to start? ,
    + Parallels: “Telephone Sol LeWitt” (SS)

  5. Jargon translation for tricks —> compare naming of surfing-snowboarding moves (beginning/middle/end) [Justin*]
    + See: parkour theory!
    + Needed: ways for people to autonomously add things w/o expert intermediary (how to limit / vet contributions? cf problems of UDict, nuisance of user login.)
    + Needed: spec for how to shoot; how to describe; glossary.
    + inverted: add classifiers for ways in which a trick is bad! for pedants, for fail-watchers

  6. 1D Logarithmic “Life” Manager [Siranush B.]

    • included logarithmic calendar allows people to see their longer term goals

    • estimates the completion based on data from other people’s calendar based on task completion

    • import protocols of work/experiment/study will include tools that allows specify different portions of it with their relevant

      • time of completion

      • importance

      • in case of the process stopping: how the project is affected vs time graph

    • 3+1 D objectification of data (for example biography of a person or company) .

  7. Indigenous Environments [Jack]

    communities want to collaborate with each other around a shared narrative of local observations related to environmental pollution, environmental degradation, and the spiritual use of sacraments.

    • want to utilize their research to challenge current industry beliefs

    • knowledge: legal precedents of industrial pollution cases

    • data: water quality monitoring, air quality monitoring, soil erosion/soil quality, cyanomonitoring/algal blooms

    • existing tools: Indigenous Observation Network (Yukon River Basin), Local Environmental Observer Network, citizen Science tooling: citsci, scistarter, cybertracker, inaturalist, Mukurtu

    • audience: separate tribes plagued with similar problems. also activists, scientists, and funding agencies that are aligned with their cause.

  8. Semantic Web3 [Beth]

    • intro at

    • 1:Describe what is possible
      Context: “Web 3.0” is the emergent system of data, content, structures and patterns continuously being generated by peer-to-peer (p2p) interactions. In the context of blockchain, these interactions are encoded in smart contracts, automated by protocols, formalized in governance mechanisms, organized in networks and infrastructures like Decentralized Autonomous Organizations (DAOs).
      - Extending the existing Semantic Web (2.o) to include these complex and dynamic interactions requires enabling a platform, and establishing shared standards and practices, for semantic mapping of p2p interactions & transactions.

      • If these interactions are expressed and recorded with continuity across use cases, disciplines, knowledge vectors, information sources etc., a rich dataset will emerge for analysis, simulation and further collaboration.

      2:Demonstrate a key example
      - The paper Liberal Radicalism: A Flexible Design For Philanthropic Matching Funds (2018) introduced 4 use cases for a quadratic voting mechanism designed to power values-based governance that blockchain developers and social impact organizations are both galvanized to test if it “actually works,” a rare incentive alignment to enable participatory governance approaches to scale IRL.
      - can’t tell “if it works” without defining metrics, parameters, variables etc.
      - defining these parameters w/ shared standards enables tracking throughout the ecosystem while also providing the possibility for many applications of simulations and modeling
      2b: Name counterexamples

      3:Define how to share results

      4:Communicate, organize, iterate
      dOrg, a consortium of researchers, developers and designers is tackling the necessary “collaboration at scale” required for generating a usable SemanticWeb3 by creating open-source tools and supporting collective action toward capturing and mapping interactions as they emerge.
      - Our first project is developing a DAO creation toolkit for DAOstack, “an operating system for collective intelligence.”
      - Collaborate with us through dOrg (Github, Riot) and the Mechanism Design Working Group (Github, Telegram).

Inception practice in groups: Imagine + illustrate an example

Tuesday (optional)

Wikipedia + Wikidata workshop in 14N-132, 3-5pm
Especially focused on public domain works as this is Public Domain Year ~!


Lightning talks

  • Gabe: PubPub (as example)

  • Beth: Enabling organizations at scale in chainspace, working with DAOstack

  • Jabari: category DBs: applying category theory to knowledge representation

  • see also: nanopubs;

Beth on orgs at scale:
+ modularizing knowledge; aligning across the diaspora
—> simple solution: naming the space, for alignment
+ Mapping intention + causality across groups
—> [Law || compare how contracts succeed or fail here] —> legal contracts as an interstitial layer of semantic authority encoding complex behavioral systems in human<> machine-readable language

Jack: many conversations around blockchain + solutions on datasets involve incentivization, whereas opensci rests on philosophy of open access
SJ: outcome-based vs. incentive-based collective action models. Incentive-based models almost always fail because the first group that manages to master the first generation of the rules gets to define the second (third+) generation of rules.
Beth: a global ecosystem of collaborators needs to be able to contribute to collective action research initiatives, which requires contributors of across disciplines and levels of technical knowledge generating and sharing data about their org’s usecase that’s semantically interoperable with internal definitions and understanding of incentives, actors, variables etc. but also translates directly to computational semantics with the eventual goal of generating an open-source dataset for simulations and generative toolkit design
SJ: game-theory vs. design-theory

Jabari on category DBs:
+ applying category theory to knowledge representation
Exs — friend-of relations can be compile into acquaintance relations; you can build larger inference rules as well.
+ Mapping — Data in other forms/dbs can be mapped into a category perspective, with a set of small libraries. functors let you move among category sets or descriptions.
+ Cf. Coq libraries that use categories (arXiv; github) | You can have confidence levels for each mapping; allow for extrapolation based on some threshhold for the composite

Pair editing of case studies



No comments here

Why not start the discussion?