Skip to content

spec.md

Latest commit

 

History

History
139 lines (105 loc) · 7.93 KB

spec.md

File metadata and controls

139 lines (105 loc) · 7.93 KB
Raw

The Context Translator — Working Spec

Hackathon track: NSD democracy for u & me How might we: match citizens to participation opportunities that are relevant to their specific context, life, and deliver that information in a way they'll actually participate.

1. Thesis

Profit-oriented tech defaults to generic solutions because that's what scales. EU democracy can't afford that default — with 24 official languages, 27 institutional cultures, ~30–40% of citizens with low digital access, and huge urban/rural/age gaps, "good enough for most" structurally excludes tens of millions of people. We should use technology to do the opposite: systematically tailor channel, register, and depth per audience segment, so that no group is excluded by design from EU democracy.

2. One-sentence solution

A bidirectional context translator: stakeholders describe what they want to ask EU citizens; the system orchestrates a multi-channel, multi-register outreach campaign (digital and physical, with NGOs, academics and local intermediaries in the loop), then aggregates responses back into a synthesis the stakeholder can act on.

3. Why bidirectional matters

Outbound-only is PR. The novelty — and the democratic value — is the return path: an elderly rural citizen writing "I want to pay less for heating" on a paper form gets weighted, summarized, and surfaced to the policy-maker alongside a LinkedIn comment from a Stockholm engineer. The infrastructure makes participation legible regardless of channel.

4. Worked example — Energy efficiency

Stakeholder: A DG ENER policy unit wants citizen input on the next Energy Performance of Buildings Directive.

The system designs five outbound channels in parallel:

Channel Audience Register Human partner
Stakeholder dashboard (intake) The stakeholder Conversational AI —
Email expert panel Academics, building scientists Formal, evidence-based National research networks
TikTok short Teens, students Fast, visual, emotional Youth NGOs (e.g. ESN, Generation Climate)
LinkedIn long-form Middle-class professionals Policy briefing EU comms team
Paper letter + prepaid return Low-digital-access citizens (rural, elderly) Plain language, 3 questions National post + local NGOs / libraries

Inbound aggregation: a synthesis dashboard converts all of these — comments, video reactions, paper ticks, focus-group transcripts — into one comparable set of themes, weighted to avoid privileging the high-digital channels.

5. Architecture (flowchart)

flowchart TD
    A["<b>Stakeholder</b><br/>NGO · MEP · Commission · Mayor"] -->|"'I want EU citizen input on<br/>energy efficiency'"| B["Semantic AI intake<br/><i>scope, urgency, depth, geography</i>"]
    B --> C["Topic + Audience analysis<br/><i>demographics, digital access,<br/>languages, civic infrastructure</i>"]
    C --> D{{"Channel Orchestration Engine"}}

    D --> E1["LinkedIn long-form"]
    D --> E2["TikTok / Instagram"]
    D --> E3["Academic expert email"]
    D --> E4["Paper letter via post"]
    D --> E5["In-person assembly"]

    E1 --> F1["Partner:<br/>EU comms team"]
    E2 --> F2["Partner:<br/>youth NGOs"]
    E3 --> F3["Partner:<br/>research networks"]
    E4 --> F4["Partner:<br/>libraries, local NGOs"]
    E5 --> F5["Partner:<br/>local councils"]

    F1 & F2 & F3 & F4 & F5 -->|"<b>Outbound:</b><br/>context-adapted message"| G(["EU citizens<br/>in their own register"])

    G -->|"<b>Inbound:</b><br/>comments, forms, voice notes,<br/>paper ticks, transcripts"| H["Inbound collection<br/>+ inbound translation"]
    H --> I["Synthesis dashboard<br/><i>weighted across channels<br/>so non-digital isn't lost</i>"]
    I --> A

    K["Open-source civic-tech consortium<br/><i>NGOs + academics, EU-grant funded</i>"] -.->|"maintains each channel adapter"| D
    K -.-> E1
    K -.-> E2
    K -.-> E3
    K -.-> E4
    K -.-> E5

    classDef stakeholder fill:#003399,stroke:#003399,color:#fff
    classDef citizens fill:#FFCC00,stroke:#b38f00,color:#000
    classDef community fill:#e8d5f2,stroke:#7a4a9e,color:#000
    classDef synthesis fill:#fff3e0,stroke:#b86b00,color:#000
    class A stakeholder
    class G citizens
    class K community
    class I synthesis
Loading

6. Differentiation

  • Decidim / CitizenLab / Consul — browser-first platforms, citizens must come to them. We meet citizens where they are.
  • EU Citizens' Panels — single-format, expensive, slow. We orchestrate many formats in parallel.
  • Targeted advertising — outbound only, optimizes for clicks. We optimize for legible democratic input and close the loop back to decision-makers.

7. Sustainability model

Each channel adapter (e.g. "low-digital-access paper letter") is an open-source civic-tech component, maintained by a consortium of NGOs and academics, funded via EU democracy / civic-tech grants — Erasmus+, the Citizens-Equality-Rights-and-Values programme, Digital Europe. The orchestration core is also open-source. This matches how Decidim and OS Politics already sustain themselves, so it is not speculative.

8. Mapping to the whiteboard problem list

Whiteboard problem How we address it
P1 No commitment / trust Closing the loop — citizens see their input synthesized back
P2 Lack of info for participation Context-adapted briefing per channel
P3 PPL don't trust the process Transparent inbound synthesis, third-party NGOs in the loop
P5 Lack of tool training No new tool to learn — they use the channel they already use
P6 Busy life 3 questions on a paper form is a 90-second commitment
P7 Lack of education Register adapts to the audience, not the other way round
P10 Accessibility Non-digital channels are first-class, not a fallback
P11 Low literacy Plain-language adapter is a maintained component, not an afterthought
P12 Multiple authentication Each channel uses its own native trust (post, NGO, email)
Hard language / hard tech The translator's whole job

9. Judging-criteria mapping

Criterion Argument
Relevance Directly attacks 9 of P1–P12 (table above)
Feasibility No new tech required — orchestration of existing channels + AI for segmentation. NGOs are already in this space.
Long-term sustainability EU-grant-funded open-source consortium per channel
Scalability Adding a new channel adapter is incremental; orchestration core scales linearly
Originality Bidirectional + multi-register + non-digital inclusion — none of the existing platforms combine these three

10. Scope of the hackathon prototype

In the demo:

  • This spec
  • The flowchart (above)
  • 6 HTML mockups under /mockups
  • 10-slide pitch deck

Explicitly out of scope (roadmap, not demo):

  • Any working ML/NLP — the AI decisions are mocked
  • Authentication, GDPR plumbing, signature collection
  • Real partner network — we name plausible partners as examples
  • Real cost figures — illustrative numbers only

11. 10-slide pitch outline (draft)

  1. Logo / title — "The Context Translator — democracy that meets every citizen where they are"
  2. Problem — generic civic tech excludes by default; show whiteboard P1–P12
  3. Solution — one-sentence + the flowchart
  4. Validation — energy-efficiency walkthrough with the 6 mockups
  5. Alternatives — Decidim, Citizens' Panels, targeted ads — and why each falls short on inclusion or return-path
  6. Viability — feasibility (no new tech) + sustainability model (grants + open-source consortium)
  7. Implementation — channel adapter as the unit of work; orchestration core as the integrating layer
  8. Next steps — pilot with one DG and 2 channels; recruit 3 channel-adapter maintainers
  9. Team — civic-tech-heavy team; emphasis on building open infrastructure, not selling a SaaS
  10. Close — "Democracy at the EU scale only works if it's polyphonic. We build the translator."