tactical direct-action decision support app
您最多选择25个主题 主题必须以字母或数字开头,可以包含连字符 (-),并且长度不得超过35个字符
Norwin 4262a76b2b foo 2 个月前
LICENSE Initial commit 2 个月前
README.md foo 2 个月前

README.md

Hivemind

Tactical Direct Action Decision Support App.

This is a first design document.

Alternative title: actfast -- fast action & response crowd planning

Goals

Provide a tool to activists that…

  • enables situational awareness / tactical decision support for mid - large scale actions
  • is low friction to foster information crowd sourcing
  • provides GIS tooling for direct action planning
  • replaces Twitter (and other platforms that are general purpose / untrusted / commercial) as action-related news source
  • is low trust & resilient to censorship or sabotage

Features

  • mark situations of interest with location
  • advertise planned actions and calls for action
  • notify about danger / call for support
  • replicate evidence footage of rights violations
  • publish public web feed?

Non-functional reqs

  • based on peer-to-peer network with multiple transports (Briar?)
  • provides some degree of anonymity and/or plausible deniability
    • trusted/untrusted nodes receive different amounts of info
    • action days can have different trust models?

Architecture

  • Tier1: Transport: Bluetooth, …?
  • Tier2: P2P Network
  • Tier3: Trust / Crypto
  • Tier4: State / Content
  • Tier5: UI

Tier 1-3 are implemented already in apps like Briar. Is this reusable? Ideally Tier 1-4 (or at least 3-4) would be implemented in a reusable SDK.

Content

All kinds of info have a timestamp, location (direction?) & TTL.

  • Infrastructure (blockades, weak spots, support centers, …)
  • Planned action (textual, public/confidential)
  • Call for action / support
  • Call for urgent help (logistics / medical / psych / legal)
  • Rights Violation (evidence)

Considerations:

  • Information trust through votes?
  • Staleness filter (UI only?)
  • Consolidation of duplicate entries (UI only?)

Trust Models

  • Different models selectable for each action day / project
  • In all models, the p2p nets’ outside environment is untrusted (IP network, radio footprinting, …)

  • Peers added in person are trusted (opportunity for key exchange!)

  • A node can vote on trust in others, advertise its own compromise

  • Confidential info (own location, planned actions) is pushed to trusted peers only