Digital Democracy 2030 Study Group
Project Summary.
Background of launch.
Press conference by Takahiro Yasuno on January 16, 2025 to launch the "Digital Democracy 2030" project
Project Objectives.
Realization of a new form of democracy using AI and digital technology
Visualization of diverse opinions, including the voiceless, and their reflection in policy
Realizing a Digital Vision of Democracy for 2030
political echoes
Response from opposition.
National Democratic Party:
Secretary General Haruha Katsuya: Mentioned Taiwan's "join-in" and considered participating in the project.
Councilor Takae Ito: Consideration of the use of the KDPJ in the policy-making process
Representative Yuichiro Tamaki: Supported as "a system that gathers the wisdom and power of all citizens to create policies and laws.
Constitutional Democratic Party of Japan:
Acting Representative Kiyomi Tsujimoto: expressed willingness to participate in the demonstration experiment.
Japan Restoration Association's response.
Representative Hirofumi Yoshimura, Governor of Osaka Prefecture:
Express support for the project
Proposed to utilize a focus on "reforms to lower social insurance premiums and social security reforms."
Willingness to use broad listening and AI to develop policy
Political Significance.
Several opposition parties officially announced their participation.
Pointed out as a possible key to regime change
Nishio comments, "If all major parties support digital democracy, Japan will move forward regardless of which party is in power."
I want the ruling party to do it against the opposition.
Evaluation of Mr. Anno's "Lagrange Point"-like strategy
GPT-4.5.icon
Lagrangian point-like strategy is "a strategy of taking a position in a situation where there are several large forces, and where these forces check each other and cannot move, or where it is difficult to move aggressively to take a position.
In physics, a "Lagrange point" is a point where gravity is in balance and a small object can exist stably. Applying this metaphorically to strategic theory, it refers to a method of finding a place where several strong competitors are in conflict or checks and balances with each other, where it is difficult for them to intervene in order to maintain their balance, and how to avoid competition and establish a stable position there.
We do not support any particular party or advocate regime change.
Main Products
Broad Listening "Broad Listening AI"
Overview:
Developed in reference to "Talk to the City
Functional improvements tailored to the practices of Japanese local governments and politicians
Using AI to analyze citizens' voices for policy making
Technical Features:
Deployment to Azure, Docker, and other infrastructure aspects
Previous versions were styled to run Python scripts at the analyst's fingertips
Reports can now be obtained by CSV upload, followed by import from Google Spreadsheet, so that results can be obtained without the need for engineering work.
Ability to visualize large numbers of opinions in a short period of time
Design to withstand large scale public comments such as 30,000 is under consideration
Large-scale deliberation support system "Idobata"
Overview:
Online platform for gathering and discussing the opinions of a large number of people
Provides a forum for large-scale discussions in which a large number of citizens can participate
Technical Features:
New functions to support discussions with AI, such as SNS-connected bots and visualization of issues
Active testing of "idobata-analyst" in action and suggestions for improvement
Demonstration.
Tokyo AI Strategy Idobata Conference" website will be open to the public (March 10, 2025).
Citizen-participatory discussion on the issues of the Tokyo AI Strategy Council
Utilized in discussions at the Tokyo AI Strategy Council, of which Yasuno is a member
Gathering citizen input on municipal AI strategies
Project Progress.
Activities for Week 1 (March 13 - 19, 2025)
Community Launch:
Set up Slack for development groups & user groups
Slack has been a success, attracting more than 250 participants in the first day after the public announcement.
A little more than 500 people as of 2025-04-11
Analytics for 5/9
https://gyazo.com/8d771103ca59b791988c16d08f69c2ec
Multiple development channels launched, including "IdoBata" and "Hiroko AI."
Diverse members including programmers, designers, PdMs, and municipal officials gathered in a self-introduction channel.
Participants with a passion for "participating in OSS development," "solving social issues with AI," "digitally updating government systems," etc. will gather.
Development of the Idobata system:
GitHub repository is opened to the public, and the initial development environment is being developed.
Progressive improvements focused on improving ease of use
Documentation makes it easier for repository users to get an overview of the project and contribution rules
Improvements begin to be made for access from a variety of devices, including mobile compatibility.
Development of a wide hearing AI:
Unification of development environment through Dockerization, consideration of features to facilitate deployment to Azure, etc.
Increased discussion on UI improvements
Community Outlook:
Expectations are high that the demonstration experiment will not only be conducted in Tokyo, but will be expanded to other regions as well.
Week 2 (March 20, 2025 - March 26, 2025) Activity
Community growth:
More participants since last week, active exchanges throughout Slack
Significant progress was made, especially in the development of "Idobata" and "Hirohyo AI".
Idobata development gains momentum:
Active discussion on creating new functions to support discussions with AI, including prototype SNS-linked bots and visualization of discussion points.
Meet and work sessions are held among developers to test "idobata-analyst" in action and make suggestions for improvement.
Progress in Broad Listening AI (Broad Listening):
Deployment to Azure, Docker, and other infrastructure improvements are proceeding at a rapid pace.
Increased opinion analysis using test environments
Progress is being made in studying the application of functions to improve the efficiency of public commenting operations (DX) at government ministries and agencies and local governments.
Cases of actual in-house and civic group pilot implementation will also be reported.
This in-house case study refers to my introduction of Cybozu.nishio.icon
Spread of new linkages:
Consultations in collaboration with several municipalities and researchers reported
How AI can be used in government and politics" begins to be considered in concrete terms.
Activity for Week 3 (March 27-April 2, 2025)
Summary of Activities:
A total of 126 messages exchanged on 9 channels
Several major development projects and event preparations take center stage
Progress in broadband AI development:
Code maintenance: consultation around formatters such as Prettier and Ruff, Makefile and Docker-related refactoring in progress
Deployment operations and data persistence issues: Issues with deploying to cloud environments such as Azure Container Apps were discussed, and the question was raised, "Wouldn't it be better to work with external storage such as S3?" and "Shouldn't we work with external storage such as S3?
nishio.iconThis is because I did First Penguin and found out that "if you try to follow OSS updates and docker build & push, the generated reports disappear because the container is overwritten", so I knew I needed to deal with it.
If docker compose up in the local environment, the local file system is mounted, so the report is not lost even with re-build.
-> You may not realize where they are stored. /server/broadlistening/pipeline/outputs/* is the individual data and server/data/report_status.json is the list
For example, if you build and deploy a container to Azure Container Apps, the generated reports are stored in the container, so when you build and deploy new source code, the behavior is that "the container is replaced by a new one = the existing reports The existing report will disappear.
-> /scripts/fetch_reports.py can save it on hand, so you can do that.
I've been running a server that I've been setting up experimentally, volume mounted in a VM.
Support for large data: "We want to design the system to withstand large scale pubic comments, such as 30,000 comments," and discussions are underway on how to improve memory usage and clustering performance of embedding.
Continued development of the Idobata system:
Major improvements to the user interface are underway.
Enhancement of discussion visualization function: Development of a function that organizes multiple opinions and categorizes and displays them by issue.
Consideration of multilingual support: Discussion of multilingual support with a view to future international expansion
Proposals for the development of a mobile app version begin to emerge.
Increased activity in functional improvements based on feedback from actual user testing
Discussions begin on the function of linking "IIDOBATA" and "Hirohiko AI".
Hackathon Preparation:
Preparations are underway for a hackathon to be held in mid-April.
Discussion on participant recruitment and venue arrangements
Activities for Week 4 (April 2 - 9, 2025)
Overall community activity:
A total of 887 messages were posted across Slack, with lively discussions centered on "Idobata" (252 messages) and "Spreading AI" (95 messages)
New channels such as "Beginner's room" and "Design_general" have been added to enhance the system for accepting new participants.
Development of the Idobata system:
PRs for new features and bug fixes merged one after another on GitHub
CSV preview functionality, improved sample data, and CI implementation to improve development efficiency are progressing.
Discussions expanded to include enhanced mobile support and pilot operation of SNS-linked bots.
Development of a wide hearing AI:
Numerous Issues/PRs closed, including Azure Blob Storage integration, formatter introduction, and cluster analysis enhancements. Windows environment support, measures to control excessive summarization, etc. are still under discussion.
Increased demand for improved clustering performance and UX improvements when handling large data
New project "Political Funding Board":
Politicalfunds" repository goes live, launching an initiative to make political fund balance reports visible.
Future development policies and collaboration plans to be discussed at regular weekend meetings
[Preparation for the Hackathon (4/12): Discussion on how to proceed with both online and offline activities, management of the reception, work rooms for each theme, etc.
Plans are underway to utilize social networking services, group photos, live streaming, etc. to attract participants.
This week's summary: New members continue to join, and a support environment for non-engineers is in place.
The community is becoming even more active with the exchange of ideas and opinions at the regular Saturday night general meetings and events every week.
Future Prospects
Transformation of the policy formation process.
Standardization of citizen participation in policy making
Comprehensive policy formation by making the voiceless visible
Rebooting Democracy with Digital Technology
Deployment in each region as well as in urban areas
Vision for 2030.
Adoption of digital democracy by all major political parties
Establish a political system that directly reflects the voices of citizens
Evolution of Democracy through AI and Digital Technology
Establishment as common ground across political parties
Conclusion.
Digital Democracy 2030" is not just a technology project, but an initiative that has the potential to redefine democracy in Japan.
Potential to attract support across party lines and become common ground across political divides
Sustainable development is expected through open source development and community building
The innovation and promise of Mr. Anno's "Lagrange Point"-like strategy
disclosing the secret of a trick
Devin is making this document in its entirety.
The reordering of the sections was indicated by looking at what was generated
The detailed bi-weekly events are published by the OSS Weekly Reporter and put on their website, and I give them the URL and let them read it in their browser. ---
This page is auto-translated from /nishio/デジタル民主主義2030勉強会 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.