Who is Shinta Nakayama?

A Japanese machine learning engineer and business owner, known online as "tokoroten". Nakayama is a maintainer of Kouchou AI, open-source software that uses AI to organize the opinions of many people and make them visible, and coordinates *Broad Listening* (政治・自治体・企業に広がるブロードリスニング), the book that gathers that experience.

How can anyone properly "listen" to the free-form opinions of thousands, or tens of thousands, of people? This is one of the people in Japan working on that question with technology.

In short

Shinta Nakayama (中山心太) is an engineer and business owner who specializes in machine learning and data analysis. Nakayama is CEO of NextInt Inc., and is known online as "tokoroten" (@tokoroten on X) [1].

Nakayama is one of the maintainers (people who coordinate and sustain development) of "Kouchou AI", open-source software that uses AI to organize the opinions of many people and make them visible [1][2]. Nakayama also coordinates the book that gathers this experience, 政治・自治体・企業に広がるブロードリスニング (Broad Listening; written by the DD2030 community, to be published by Impress) [3].

Connection to Plurality

Nakayama does not appear in the book Plurality. The connection runs through Kouchou AI. Chapter 5-4 of the book gives Talk to the City (TTTC) as an example of "broad listening", listening to many voices. Kouchou AI is its successor in Japan. → Talk to the City, Augmented Deliberation

Maintainer of Kouchou AI. Nakayama takes part in Digital Democracy 2030 (DD2030) and serves as a maintainer of Kouchou AI [1][2]. In slides explaining the technology in June 2025, Nakayama described the processing pipeline inherited from TTTC: extract opinions, turn them into numerical vectors, place similar ones near each other, divide them into groups, and have AI name the groups. The slides state its purpose as making visible not the quantity of opinions but their diversity, including minority opinions [4]. In July 2026 Nakayama released a version that runs the same analysis entirely inside the browser [5].

Coordinator of the Broad Listening book. The description on the book's manuscript repository (GitHub) says that @tokoroten is in charge of the overall structure and progress management. The preface is by Takahiro Anno, the endorsement by Audrey Tang, and the license is CC BY-NC 4.0 (usable for non-commercial purposes with attribution) [3]. According to August 2026 slides, Nakayama wrote about 60% and Hirokazu Nishio about 15%, with many other authors as well [2].

The original of Tang's endorsement reads: "Broad Listening puts AI in the human loop: Democracy learned to broadcast — Japan shows it how to receive." The book carries a Japanese translation in which Nishio fills out the wording: broad listening builds AI into the loop of human deliberation and gives democracy, which has leaned toward sending messages, the capacity to listen at scale, and Japan is showing a practical answer to how [6].

Digital Democracy Summit 2026. On August 2, 2026, Nakayama gave a talk with Nishio on "the technology of broad listening" [1][2].

Distance from Team Mirai. In February 2026, Nakayama wrote on X that Team Mirai no longer uses TTTC or its successor Kouchou AI, and that Polis has added AI summaries in its new version. It was a reservation about commentary that was out of date [7]. It is one clue for keeping Kouchou AI (developed by DD2030) separate from the systems Team Mirai now uses.

Links with Mitou

In the second half of fiscal 2006, Nakayama was selected for IPA's Mitou Youth program as chief creator of "Development of TsukikageSystem, a next-generation Greasemonkey" (under PM Michiaki Yasumura) [8]. In a 2011 contribution to the journal 情報処理 (IPSJ Magazine), Nakayama wrote that what makes Mitou interesting is that people who are good at different things come together naturally, and said that the essence of programming is communication [9].

Main ideas

Getting past the trade-off between scale and depth with AI. Nakayama explains broad listening in a single slide. With a fixed budget, asking many people means you can only ask simple things, and asking in depth means you can only ask a few people. Broad listening, Nakayama says, is getting past this either-or with the help of AI. As an example, the slide notes that the 2025 national census, which asked every household 17 questions, cost about 94.6 billion yen [2]. → Broad listening

Three forms of broad listening. The same slides divide broad listening into three [2].

Broad listening as a tool — organizing large amounts of free-form writing with AI, as TTTC and Kouchou AI do
Broad listening that supports augmented deliberation — AI entering into discussions between people, as in Idobata
Broad listening led by politicians or government — the efforts of parties and local governments

The slides sum up that the second form spread overseas, and the third in Japan [2].

Turning the anger of online outrage into negotiable proposals. In a long post on X in March 2026, Nakayama introduced the political scientists W. Lance Bennett and Alexandra Segerberg's distinction between "Collective Action" (movements of organizations) and "Connective Action" (movements of connection). Connective action, such as outrage on social media, has no representatives to negotiate, Nakayama argues, and so has trouble reaching as far as changing institutions. What is needed, then, is a mechanism that turns the anger born in such outrage into proposals that can be negotiated, and connects them to official channels; that, the post proposes, is exactly the role of broad listening. Nakayama notes that the latter part is a personal framing and proposal, not something academically established [10].

Wariness of filtering. During the event Funding the Commons Tokyo in July 2024, Nakayama wrote on X, in substance, that filtering out aggressive remarks before broad listening is very good, but it also sits right next to a society in which aggressive people lose their social rights [11].

Uses in companies. As part of DD2030, Nakayama interviewed Altius Link, Inc., which uses Kouchou AI to analyze the voice of the customer in its contact centers (interview on December 9, 2025) [12].

Sources

Digital Democracy Summit 2026, session page for "ブロードリスニングの技術" (in Japanese) — https://summit.dd2030.org/sessions/broad-listening-technology/
Shinta Nakayama and Hirokazu Nishio, "ブロードリスニングの技術", slides (2026-08-02, in Japanese) — https://www.docswell.com/s/tokoroten/5Q2LVW-tech-of-broadlistening-dd-summit2026
README of the manuscript repository of Broad Listening — https://github.com/digitaldemocracy2030/broad-listening-book ［CC BY-NC 4.0］
Slides explaining the technology of Kouchou AI (2025-06-14, in Japanese) — https://www.docswell.com/s/tokoroten/ZL1M88-2025-06-14-014546
kouchou-ai-serverless README — https://github.com/tokoroten/kouchou-ai-serverless
Broad Listening, "Words Audrey Tang Sent for This Book" (in Japanese; English translation in the same repository) — https://github.com/digitaldemocracy2030/broad-listening-book/blob/f69c60c838a91b91668abf109d7093b240610139/en/00_endorsement_from_audrey_tang.md ［CC BY-NC 4.0］
Shinta Nakayama, post on X (2026-02-07, in Japanese) — https://x.com/tokoroten/status/2020004238718570604
IPA, second half of fiscal 2006 Mitou Youth selection results (in Japanese) — https://www.ipa.go.jp/jinzai/mitou/esp/2006/2006-2-youth-koubokekka.html
Shinta Nakayama, "プログラマじゃないから続くコミュニティ" (A community that lasts because it is not only programmers) (情報処理 Vol. 52 No. 12, 2011, in Japanese) — https://ipsj.ixsq.nii.ac.jp/records/78371
Shinta Nakayama, "Collective ActionとConnective Action" (X, 2026-03-02, in Japanese) — https://x.com/tokoroten/status/2028388031192035560
Shinta Nakayama, post on X (2024-07-24, in Japanese) — https://x.com/tokoroten/status/1816010940988678391
DD2030, "広聴AIのコンタクトセンターでの活用 - アルティウスリンク株式会社に取材！" (Using Kouchou AI in contact centers: an interview with Altius Link) (note, 2026-02-24, in Japanese) — https://note.com/dd2030/n/n6db28a76ca95

We welcome error reports and questions ("What is…?"). Submissions become public issues.

Report an error in this article ・ Ask a question