POSE: PHASE II: The Free Motion Capture (FreeMoCap) Project

Introduction (Gap/Need/Hero statements)¶

Current model not fit for present and future needs
Need alternative models, organizations, infrastructures for science and education
- Generative and self sustaining
  - Von Neumann machine - Sustainibility through replicability
- As Autonomous As Possible (allow the fewest to do the most)
- Topic/software/hardware agnostic
- Multi-scale adaptable, extensible, maintainable, and replicable [SEP]
Two Loops theory of social progress
- Not saying we should burn it all to the ground
  - It’s already burning
- We need to build a better ship
  - & f-ing ASAP plz thx

Intellectual Merit (How we advance knowledge)¶

Enablabling tech-forward integrative research
Rosetta Stones and Systems Engineers
Training nextgen technowizards and cross-disciplinarians

Broader Impacts¶

Meager success:
- New useful tool
- New body of research realted to human/animation perceptuomotor control
- Education tools and strategy
Expected outcomes:
- Support and sustain the growing community of freemocap users, loosely organized around a shared love of the measurement, investigation, analysis, and expression of human and animal movement (broadly construed)
- Sustainable organization and vibrant ecosystem existing as a proof-of-concept alternative organizational approach to broadscale academic research, education, and training
Idealist future:
- Iterative solutions on ‘project-scale’ strategy produces a vibrant ecosystem of ecosystems which fundamentally changes the face of education and scientific research.

Context of the OSE¶

Current Context¶

Technogical context¶

Capacities
- Cutting Edge (Established methods)
- Bleeding Edge (Promising advances)
- Likely futures
Related projects
- Michael Black
- R. James Cotton
- MMPose
- Deeplabcut
- etc

Corporate mistrust
Privacy violations
Worker/social exploitation
Economic Disparity -> Access Disparity (and many many other disparities)

Ethical context¶

Risk of future corporate conglomoration of technology
Need for decentralized systems
- Clowdstrike outages
- Openpose dead-project status
Rise of AI
- Likelihood of future violations
- Need for fast development of FOSS tools before corpo tech dominates
Need for decentralized user-directed FOSS projects
- With accompanying organizational structures to support
Inaccessiblity of science and technogy
- Growing Gaps
  - (See NSF Strategic plans)

The problem of universities¶

Exclusionary by design
Ever hungry corporate engine driven by student debt
- Tuition, room, and board deeply immoral and malignant
- ~60% Overhead charges
‘Higher Education as a filter’ \cite{arrow1970}
- “Purpose of education is to let purchasers of labor know the value prop of a student”
  - 🤢 🤮
The ‘banking model’ of education (perjorative) \cite{friere1967}
- Education as a transactional process of certification
Current systems of academic research and education:
- ...does not reward necessary activites to build and sustain an OSE
- ...is not equipped to address present and rising societal needs and challenges

On gardens and cottages¶

Need for interoperable research and communication methods,
- ‘Cottage Industry’ model is siloed, inefficient,
- ‘Garden Ecosystem’ for development of self-supportive ecosystem (like 3-sisters from Braiding Sweetgrass)
The problem of publication (Dan Goodman stuff)
The story of Grease Pencil

Objectives and Long term vision¶

Aspirational Goals¶

Make FreeMoCap the best motion capture option available at any price
- Bench mark agaist modern gold-standard techniques and proprietary softwares [Validation], [Diagnostics]
- “We define the floor”
Universal access, all levels accessible and performant
- “We follow a “Universal Design” development philosophy, with the goal of creating a system that serves the needs of a professional research scientist while remaining intuitive to a 13-year-old with no technical training and no outside assistance.” (freemocap.org)
Community co-ownership and self-governance
- [BDFaSPoT]
Covert Education
- [Anti-paradox of active user]
- "Using the thing teaches you how it works
- [WYLTKM]-Bot

Guiding Principles¶

Universal Design¶

“When anyone can use it, everyone benefits”
Same tool for 0XP as for professional
“The [aspirational goal] of creating a system that serves the needs of a professional research scientist while remaining intuitive to a 13-year-old with no technical training and no outside assistance”
Sustainability through accessibility
- The key to sustainability is a growing, active, diverse, and vibrant userbase
- Universal design attracts the widest possible userbase

Universal Access¶

Maximize access to maximize impact
- Free software, so econimically maximum possible userbase
- Quick as possible route from “Just heard of this” to “Oh look, a skeleton!”
  - minimize ‘time to skelly’ [TTS]
  - Inside out education
    - Greased rail to the good stuff
    - Educational routes from their towards whichever sub-field draws the user [Plot-hook-strategy]
Education focus
- Core design built around:
  - Progressive Disclosure [WYLTKM]-bot
  - Modern best practices
  - Empirically measured user workflow data
    - [SkellyBot-Local]
- to avoid [paradox of the active user]
  - [User-State-Model-Trajectory-Control]
- foster long term user relationships []
- encourage community engagement
  - Push user-student into ‘zone of proximal development’
  - Provide easy access to resources (AI, static, community)
- (see AI Simulacrum Pseudo Mentorship[AiSPM])
- Covert education:
  - Using the thing should teach you how it works
  - [ZoPD] & [WYLTKM]-bot
Maintain a True Minimum
- Hardware
  - Understand and maintain an option to run with the barest, most generic, and lowest-possible cost hardware,
  - Focus and maintain a [generic hardware] pipeline as the core path
    - Specialized tools (e.g. sterocam, eye trackiner, emg, forceplates) and proprietary options (e.g. iPhone, Basler/FLIR cameras, etc) added via plug-in system
    - Developed by FMC-C and/or upon [client] request
- Software
  - Maintain and understand requirements for different outcomes (specific to defined use cases)
    - Minimal requirements (to get a viable output)
    - Recommended requirements
      - Default default (for first-timers)
      - Per use case
        Max speed (e.g. real-time, long, high-res, high-fps recordings)
        Max accuracy (e.g. scientists, clincians, artistic/atheltic performance)
    - Maximum requirements
      - Specs beyond which you won’t see an improvement in results
      - Determined by understand technical bottlenecks
  - (see Evaluation/Diagnostics)
- Cost
  - Methods to avoid price-barrier for paid aspects of [FMC-F] Activities (see #no artificial scarcity, # revenue streams)
  - [HumbleBundleModel] - Sliding scale pay options
    - Recommended cost relative to break-even.
    - Give user option to allocate overage to different programs/departments
      - (by necessity, the recommendations will be followed if at all possible, adjustments may be made to ensure long term stability of the FMC-Org.
      - Its kinda like a voting system, but with money)
      - Support for:
        New features
        Specific projects
        Scholarship/community grants program
        Unspecified
    - Users requesting at-or-below cost must submit a request, which can be evaluated and supported via the ‘communal support’ scholarship/money bucket [CommunityGrantSystem]

No artificial scarcity¶

“Knowledge is free, labor is unbelievable expensive”
Anything that can be duplicated infinitely is free for everyone
- Software
- Documentation
- Static documentation/tutorial/educational material, etc
Anything that is finite (or costs us money) can be charged, e.g.:
- Human labor
  - Service models
    - Setup/planning
    - Dedicated technical support
  - Custom Software
- Cloud Services (SkellyCloud)
  - Offer ‘free’ tier, tie to open finances, allow community to set primary constants relative to a fixed % of the [FMC-F] annual budget
  - Server side processing
    - Mocap data
    - AI LLM
- Cloud storage
- Camera hardware

Community Focus¶

Emphasize that the core of the FMC-P is the community of users
The software artifacts are the fruit of the garden
Develop lateral connections, encourage growth, exploration, community support

Aggressively open source¶

“The open source community is what the scientific community pretends to be”
FOSS as a matter of moral imperative centered on the premises of Univerisal Design and Universal Access
- Identification of strategic advantage within the contexts of FOSS being the ‘right thing to do’
- See “why open source misses the point”
Encourage conversations around moral and ethical contexts of current context of the project and society at large
Morality without Zealotry
- We have our conversations clearly, but do not exclude those who misalign provided they aren’t operating in opposition to our values
- More about being aware, explicit, and intentional around moral and ethical decisions
- Encourage and allow non-FOSS aligned activities through the For-Profit entity via B2B connections with external and spin-off entitities
  - Charge according to degree of mis-alignment with core values

The FreeMoCap Project (FMC)¶

Table 1 - (Items below crossed with current, in development, planned)

Artifacts¶

FreeMoCap Software(s)¶

Core FreeMoCap Software(s) (FMC-Core)
- Fully integrated FOSS platform for Laser Skeleton research/recording
  - Fully-body mocap
    - realtime, animator and/or science focus
  - Face mesh tracking
  - 3d gaze and eye tracking
    - Webcam eye tracker
    - Plugin support for Pupil labs
    - VR/HMD support
  - Integrated laser skeleton
    - Matthis papers
  - Analysis and visualization suite
    - Blender, threeJS, R-studio support, skellymetrics reports
- Can also serve as a generic Computer Vision platform
  - via recombination of other plugins around skellycam package
- Composed of various sub-skelly repos, each of which:
  - Can operate in a standalone format or as plug-in to core software
  - Matches general architecture and CI/CD structure of parent freemocap software, including:
    - Standalone GUI/CLI
    - .github/workflows/: CI/CD via GH Actions, including testing, diagnostics, etc
    - README.md: Minimally must including ##H2 headings for Description, Installation, Useage
    - api: FastAPI endpoints
    - docs: A documentation folder.
    - examples: Example code and scripts for standard use cases
    - templates: Template for add-on/plug-in
  - Notables:
    - SkellyCam (cameras)
    - SkellyTracker (2d tracking)
    - SkellyForge (3d reconstruction)
    - SkellyViewer (Viusualization and Observability)
    - SkellyBlender (Animation output via Blender addon)
    - SkellyMetrics (Analysis pipelines and ‘report’ building)
    - SkellyBot (AI and LLM support)

Documentation and Educational material¶

Providing best-available educational materials, tutorials, and demos for related areas (link to existing educational material for general topics, focus on our specific use case)
- Tech (e.g. Computer vision, Cameras, CNNs, etc)
- Math (e.g. Linear algebra, timeseries analysis, computational geometry, kinematic/ kinetic analysis)
- Visualization (e.g. data visualization, animation, video game development, etc)
- Science (e.g. human/animal movement, perception/action, etc )
Static Docs
- freemocap.org/documentation
- Tutorials
- Recorded lectures and live streams
AI-assisted documentation
- SkellyBot
  - Lives on freemocap.org chat
  - Lives in Community server (summonable, requests consent before interacting with new users)
  - Lives in local app (either connect to existing API via API key, run a local model, or use the [FMC-F] SkellyModel endpoint (requires account))
  - See AI-assisted-psuedo-mentorship

Datasets and Derived Models¶

See `[FMC-F]/Revenue/Federated and co-op datasets and models

The FreeMoCap Community (FMC-C)¶

Broad and deep user base, integrating
- User/Developers
- Scientist/Artists
- Student/Expert
Develop active social community space where users can share progress and ask questions
- Currently - Discord server
- Eventually moving to Matrix/Element server w/ Discord/Slack bridge
Nurture long term user support, actively encourage growth via semi-automated interactions with software and community spaces
The FreeMoCap Community userbase comprises the following heavily overlapping sub-communities:
- Researcher community: Researchers centered around integerated ‘Laser Skeleton’ research (Gestalt of vision + motor is greater than the sum of its parts), including sub-specialties of theory-driven and applied/clinical investigations.
- Developer community: Software developers and technologists dedicated to the long term stability and development of the main artifactual vehicles of the FMC-OSE
- User community of technically and geographically diverse set of end user communities dedicated to the specialized domains and cross-disciplinary bridges asscociated with the broad domain of topics, fields, and methods associated with the FMC-P.
- Metrics - Prism/Pokemon power chart of community composition, noting specialists relative to generalists
  - tie to Working in Public ‘toys’, ‘stadiums’, ‘clubs’, ‘federations’ (goal is federations)
  - We are a ‘stadium’ that wants to be a stratified federation (i.e. federation with XP Levels)
- ** Types/motivations of Users**
  - Students: Want to learn
  - Artists: Want to create
  - Scientists: Want to understand
  - Clinicians: Want to heal
  - Engineers: Want to build
  - Athletes/Performers: Want to perform
- Intellectual Content Creators
  - Software developer
    - Core
    - Visualization
    - Analysis
  - Creator Users
    - Artistic
      - Animation
      - Video Games
      - Performance (dance, athletics)
  - Educator Users
    - Science teachers
      - Scientific and clinical underpinnnings
      - Physics, computational geometry
    - Tech teachers
      - Camera technology
      - Computer vision (ML-based and classical)
      - Applied technologies (motion capture, eye tracking)
    - Coaches, trainers, physical therapists
      - Use to teach the recorded patient about their movement to improver performance assist in a clinical rehabilitation, assessment, or diagnosis.
  - Community helpers
    - Moderation
    - Troubleshooting and technical support
    - Encouraging newcomers
- OS Community building components:
  - recruitment
  - motivation
  - mentoring
  - managing
  - mediating disputes

The FreeMoCap Foundation ([FMC-F])¶

Economically self-sustaining combination research/software development institute

Organization¶

Non-profit (primary) organization handles ‘charitable activities’
- FreeMoCap Foundation, Inc.
  - Federally recognized 501c3 IRS certified public charity
  - Incoporated as non-profit corporation in Massachusetts
  - Board:
    - President, Clerk, Treasurer
  - C-Suite:
    - CEO, COO, CTO, CFO
- Follow bylaws, mission, etc laid out in Articles of Incorporation
For-profit organization handles other activitites
- Handles anything not covered by IRS definition of Charitable Activities
- SkellyTech, LLC.
  - Will convert to C-corp as appropriate

Governance¶

Crib numpy’s notes
Follow Open Finances and Open Everything procedures (e.g. Backyard Brains)
- “Sustainability through trust, trust through transparency”
Benevolent Dictator for a Set Period of Time (BDFaSPoT) Model
- Like the classic Benevolent Dictator for Life(BDFL) model, but transitions to community-DAO style management over multi-year, tailorable timespan
  - Retains flexibility and adaptability needed in a projects’ early stage, before a cohesive group identity and organizing infrastructure has developed
  - Allows controlled transition plan to more community-driven, decentralized model
Transition plan based on relative weighting of Board vs Community voting on key elements (see below)
- Present: 100% Board (No operating mechanisms that allow for direct community input)
- Eventual: 50% Board, 50% Community ([FMC-F] President is tie-breaker)
  - In case of true deadlock FMC-C can Hard Fork, but lets hope it doesn’t come to that
Staged transition to reliance on Community guidance to set various core computational constants (of increasing risk)
- Community collaboration spotlight winners (Eurovision style voting of ‘members’, i.e. people with >0XP)
  - Develop and test basic mechanisms of community voting
    - Associated management of communication, moderation, etc
  - Timeline: Immediate
- Decisions related to [Skelly Enhancement Proposals]
  - Touches core software artifacts
  - Timeline: Early proposal period
- Community Free tier for cloud services
  - Touches organization finances
  - Timeline: Late proposal period (optimistic)
- (Later) Salary Base Blocks, overhead rates, Co-op dataset/model management
  - Touches coreorganization operations
  - Timeline: Beyond proposal scope

Responsibilities:¶

Support and development for core software(s)
Support Research in related fields (including training, mentorship, and grant support)
- Neuroscience (human and non-human)
  - Motor control (mocap)
  - Visual/oculomotor neuroscience (eye tracking)
  - Developmental studies
- Biomechanics
- Applied clinical research
  - Injury rehabilitation
  - Gait and posture
  - Age-related
  - Developmental
- Robotics
- etc
Manage annual workshop and conference
- FreeMoCamp is like summer camp, hackathon, workshop, education/training focus
- FreeMoCon more of a conference vibe
- Pair them up, so FreeMoCamp runs during the week and leads straight into FreeMoCon on the weekend
Handle security, privacy of data, codebase, and user info
Pay salaries of a core set of maintainer/developers
Offer contract work to the FreeMoCap Community
Manage finances
- see (Revenue Streams)
- Things covered under ‘charitable activities’ IRS definition are handled by [FMC-F]
- Non-charitable activities handled by subsidiary for-profit (Skellytech)

Licensing Model¶

AGPLv3 for all repos owned by [FMC-F]oundation
- “Its a dart frog party, and everyone is invited!”
- AGPLv3 as protection from exploitation by proprietary softwares
- reference ‘open source misses the point’
Exclusive sub-licensing rights to SkellyTech
- SkellyTech handles B2B license agreement, whereby we offer software using non-AGPLv3 license
  - Cost dependent on ‘distance’ from AGPLv3 (AGPL -> GPL -> LGPL -> Apache/MIT -> ... -> Proprietary)
- Can offer sub-license agreement for existing software, or to create custom software solutions

Proposed actvitities¶

Activities to bring us from current status towards the Ideal Perfect status described above

Ecosystem establishment and Growth¶

User Engagement/Education¶

Model User as a dynamic trajectory though the >3d space with axes:
- User vs Developer
- Scientist vs Artist
- Novice vs Expert
- also - any definable vector
  - Mentorship vector
    - Defined as another user’s UserStateVector (USV)
  - Community vector
    - Defined as median USV of users within a given community (with XP level > threshold)
- Use rigid milestones, stochastic user-matching, and AI-analysis to determine each users’s ‘Status’ relative to each sub-topic, sub-skill, sub-community:
  - Proficiency [P]
  - Interest [I]
  - Experience [E]
- Mis-matches between P, I define ‘gentle pressure’ vectors
- Constant gentle pressure to increase E via:
  - Using more advanced methods
  - Gamifying improvement metric outputs (see validation and diagnostics)
  - Diving deeper into technical and scientific underpinnnings
- PID controller with stochastic noise encouraging exploration into ‘zones of proximal development’ (ZoPD)

AI-mediated pseudo mentorship support (SkellyBot)¶

AI modeled to provide a SIMULATION of real human support
Explicitly AI at all times, intended to provide infinitely duplicable simulation of replicable text based emotional labor
100% transparancy to user re:
- Data available
- Derived information (i.e. UserStatus)
- Computations, processes, prompts, etc
Offer user direct control (i.e. they can explicity alter and edit their core data model to generate different support behavior from the AI)
- e.g. let them directly tell the machine their interests, expertise, etc
SkellyBot sub-personas:
- Onboarding-Bot
  - Goals:
    - Primary: Introduce new users to software
      - Onboarding tutorials
      - WYLTKM-bot
      - RTFM-bot
    - Secondary: Encourage engagement with community
      - Community Server spaces
      - Community Activitities
      - FreeMoCamp/Con
- ReadTheManualBot (RTFM-bot)
  - Goals:
    - Primary: Help them learn and run the software
    - Secondary: Avoid ‘paradox of active user’ (offer quicker paths to victory)
  - RAG model
    - Docs (FreeMoCap Technical Docs)
    - Logs (User current and past logs, Github Issues, Server Help Requests channel)
    - Images (e.g. from cameras, videos, or screenshots)
    - Data (npy, csv, sql, json, with helper readers)
- WouldYouLikeToKnowMoreBot (WYLTKM-Bot)
  - RAG model
    - Docs (FreeMoCap Educational Docs, blog, YT transcripts, Server chat records)
    - Wikipedia
    - Search Engine Request (SerPAPI)

Community Building¶

Annual Workshop/Conference: FreeMoCamp/Con 💀⛺¶

FreeMoCamp: workshop/hackathon/training focus
FreeMoCon: Confrence style, talks, showing off
Inspirations:
- Neuromatch (for virtual component)
- Cold spring harbor (short term research project/workshop timeline)
- Get an example of a hackathon from the Endurance crew
In person & Virtual
- Held at YMCA summer camp, or similar
  - All inclusive ticket - lodging and food handled by summer camp facilities
- Online component via dedicated Element space (with Slack/Discord bridges)
- ANYONE can view all online materials, and follow along activities (encouraging conversationin public server)
- Paid online participants may engage directly in the dedicated event space, and will get direct engagement with folks at the in-person event (for troubleshooting/planning/etc)
- Pricing
  - Corpo pricing -> 100% standard price (Set high enough to make the lowest pay tiers make sense)
  - Academic pricing -> 50% standard price
  - Community members (defined by badges and community participation) -> 25% Standard price
  - Additional STEEP discounts and scholarship programs for students, educators, researchers, artists, etc
    - Minimum price w/o scholarship = break-even price per participant
- Combination hackathon, training workshop, and scientific conference on ‘fully body movement research’
Participants in teams
- Plan and collect a data set
- Analyze and present results
- Encourage collaboration and data sharing between teams
- Offer training/presentation ‘tracks’ for
  - Data collection
    - Understand and use the core software
  - Data analysis
    - Analyze and computationally model data
  - Data visualization (and animation/art)
    - Create visualizations and animations from freemocap data

Community Challenges¶

Offer a prompt for a kind of movement/activity to record
- Community works to record the best entry, submitted via Discord/Element/Freemocap.org
  - Community votes in ‘Eurovision style’
    - i.e. each member has a fixed number of voting points to distribute among entries
    - Maybe Badges provide additional voting points?

Gamified documentation/software training¶

Offer Badges and stuff in discord for milestones, i.e.
- First contact - open the software
- First record - Anything recorded
- First success - Successful recording
- e.g. also calibration, data analysis, PR, etc
Apply badges in Discord/Element server (associated with a freemocap.org account?)
- Badges offer ability to send ‘top level’ messages to otherwise view-only channels (i.e technical spaces)
  - Anyone can create a thread from a top-level message though
  - The idea would be to create spaces where higher XP folk can talk about complicated stuff without distraction [WorkingInPublic]
  - But also LowerXP folk can still see the conversations and ask questions (Maybe make it easy to summon SkellyBot to explain a message/conversation 🤔)

Organization and Governance¶

Build administrative infrastructure
- Financial planning and forecasting
- Legal definitions, B2B contracts, Co-op Datasets & Model ownership and dividend payouts.
Developing Standard Operating Procedures
- Operational
- Organizational
Establish Core maintainer actvitites
- Build team from relevant expertise
Develop Skelly Enhancement Proposal system

Continuous Development, Integration, Evaluation Model¶

Development¶

Core:
- Architecture
- Functionality
- Observability
- Usability
- Accessibility
Bug Fixing
- GH Issues and Comminity server
- Encourage newcomers to assist with:
  - Triage
  - RTFM requests
  - Sleuthing
Roadmaps
- Skelly Enhancement Proposals (SEP’s)
  - Like PEP system
  - Post in public forum (GH Discussion board? Bridge/sync with Community space)
  - Topics
    - Core
    - Feature request
    - Quality of Life request
    - Protocol change request (i.e. style guide, naming conventions)
    - Data model change
Add-on support
- Plug-in system
  - Alternative to an existing core sub-skelly package
    - Must interoperate with existing architecture invisibly
  - Additional add-on extending core functionality
    - Extends functionality from existing API
- Templates provided per-sub-skelly-repo
  - Maintain consistent style-guide and consistent naming standards

Integration¶

Automated versioning/release cycle
Git flow version control model CONTRIBUTING.md
Backwards compatibility, deprecation and Long Term Support (LTS) versions

Evalution¶

Automated Testing framework¶

Quick tests on commit to PR
Full tests on merge to main
Testing for:
- stability (unit tests)
- security (e.g. dependabot),
- replicability (e.g. test datasets)

Validation framwork¶

Is the core tool providing accurate measurements of world?
External Validation
- Against ‘Gold Standard’ marker-based mocap (e.g. Qualisys)
- Ideally using standard mocap validation techniques from clinical biomechanics (ISB?)
Internal Validation
- Validation within a given dataset via error residuals,i.e.
  - Reprojection error (2d tracker validation)
  - Rigid body correction residuals (3d data validation)
Provide tutorials for running internal validation studies, providing both onboarding training, providing data/user confidence, and optionally getting them to offer their data/results to a ‘community validation’ datapool (see also, (Fed/Coop Datasets and Models))

Diagnostic framework¶

Automated system to measure perforamnce (speed, stability) on standard hardware/task benchmarks
- Involves the standard test pipelines, but with the scale of the test data pumped up
Minimal requirements to maximize accessibility
- Understanding and clearly defining ‘minimum specs’ to achieve viable output
- Strive to decrease these minimums while remining mindful of trade-offs
- Minimum reqs also define max speed, relevant to realtime applications
High-water marks and benchmarks
- What is the best performance possible with unconstrained time and resources?
  - Speed (real-time processing)
    - FPS on standardized PCs at Min, Rec, Max setttings
  - Accuracy (scientific applications)
    - Validation score on Min, Rec, Max settings

Sustainability¶

Definitions, Strategies, Goals, and Metrics¶

Community growth metrics¶

Raw numbers (more users is good)
Community make-up
- Topics of interest
- Expertise and background
- Demographics
  - Young appeal - Good for education and outreach goals
  - Older reach - Good for professional appeal/software quality
  - Geography - Global distribution and spread should match popultion density
  - Racial/economic/gender distributions should match population statistics
Individual user trajectories
- Minimize ‘bounce’ (Single pings)
- Maximize ‘velocity’ (extracted user growth non-stagnant, e.g. anti-paradox-of-active-user protocols)

Organizational stabilty metrics¶

Ten-year plan milestone:
- Calculate forecasts based on various optimistic/realistic/pessimistic parameters
- MILESTONE(S): 10 year operational window at each level

Sustainability plan: a virtue ethics approach¶

“Sustainability through quality”¶

Generic validation metrics should be strictly monotonically increasing
- i.e. global measure of ‘goodness’ of mocap data should ALWAYS increase over time (see Validation)

“Sustainability through replicability”¶

Maintain organizational documents, Standard Operating Procedures (SOPs), and eductional outreach to encourage others to create organizations with similar stucture to [FMC-F]
Good for other people from wildly different domains (not relevant to [FMC-F] mission)
Could spawn sub-orgs from with [FMC-F] (optionally existing within [FMC-F] umbrella, either as non-profit or for-profit org)

Sustainability through revenue¶

Other big FOSS orgs - Patchwork financials

Revenue streams¶

Federated/Co-op models and datasets
- Federated datasets
  - e.g. Users able to ‘donate’ their data, while retaining ownership (earning divdends)
- ML/AI model(s)
  - Train our own CNN Tracker and LLM via ‘Federated learning’ scheme
  - Architecture allows any user to train/fine-tune the FMC Base Models using their private data, OPTIONALLY contribute updated model weights to the FMC-Org (e.g. see Flower)
For Profit sub-entity
- Skellytech, LLC
  - B2B connections and sub-licensing of:
    - Core-software
    - Derivative models (co-op data ownership)
  - Any arrangement that does not fall under the ‘charitable activities’ allowed by 501c3 guidelines
- Spin-off companies (FreeMoCap as tech incubator)
  - Incubator support for FMCF member/students who want to start businesses or organizations based on freemocap related technology and tools
  - Offer umbrella support (we support them with our org infrastructure, e.g. grant support), or
  - Launchpad support (pathway to independence)
  - Offer exit (any time) and re-entry (if possoible) mechanisms
Manage grants, both for org (e.g. NSF POSE) and individual research grants for members of FMCF (e.g. R01’s and F32’s)
- Offering extremly low overhead costs, relative to universities. e.g. 10-15%
Selling hardware/cameras
- first, dropship standard cameras and packages (low-to-high quality)
- later, develop custom cameras (including education focused cameras)
Donation drives
- Direct donation pushes
- Ask for addon donation on checkout when purchasing other stuff (Humble Bundle model)
Clients (Contracting from FMC community)
- Custom code/software
- Project planning and consultation
- Technical support
- Direct teaching/training
Annual Workshop (FreeMoCamp)
Annual Conference (FreeMoCon)
Server-side processing
- Processing users’ data using high-end hardware (more than they are likely to have available)
- AI/LLM application useage
Cloud data storage

Security and Privacy¶

Fully offline local processing option
- Users can always run totally dark in no-internet environments
Google Cloud Platform for secrets management
- Following industry standard based on Google Cloud Platform recommendations
Matrix homeserver for privacy, security, consistency
- Affiliate with large and vibrant open source community around the Matrix specification
  - Let them figure out the encryption/security/privacy stuff
Full instruction set on how to duplicate technical and organizational infrastructure with documentation and tutorials
- Promotes replication/duplication by aligned parties
- Invites discussion, critique, analysis, and collaborative improvement
- Build federated network of operationally interoperable organizations

Notes

Official Resources

Document Notes

Document Formatting Notes