Code-Driven Manifesto: Auditing Japan's Public Data & Systems from an Engineer's Lens

Hey — okamu here! Today I dig into how Japan publishes government data and what engineers can actually do with it.

• Japan is building dashboards and standardization efforts (Digital Agency, Japan Dashboard, e-Stat), but machine-readability and API uniformity are uneven.
• Many useful numbers are visible on dashboards, yet often locked in PDF or proprietary widgets — turning them into reusable data needs engineering work.
• Practical fixes: publish CSV/JSON + DCAT metadata, a shared open API spec for municipalities, Git-driven publication + CI for data quality.

結論

Public-sector data in Japan is moving in the right direction: Japan Dashboard, e-Stat and Digital Agency initiatives provide centralized visibility. エンジニア的に言うと、ここはAPI一本とスキーマ一つで劇的に改善する話なんですよ。今ある取り組みを“見せる”から“使わせる”に変えるのがキモです。

Report: what I looked at and what I found

Sources checked:

• Japan Dashboard (Digital Agency): https://www.digital.go.jp/resources/japandashboard
• e-Stat visual dashboard and API: https://dashboard.e-stat.go.jp/ and https://www.e-stat.go.jp/
• Digital Agency policy page on local governments: https://www.digital.go.jp/policies/local_governments
• GovTech Tokyo case pages: https://www.govtechtokyo.or.jp/services/data-utilization/

Key findings:

1) Format fragmentation

- Dashboards are great to read, but many underlying datasets are embedded in PDFs or siloed widgets. 要するに、スクレイピング前提になってるデータがまだ多いということです。

2) API coverage & quality

- e-Stat has an API (good!), but coverage across ministries and local governments is inconsistent. Some municipalities still publish only HTML/PDF on gc.jp/lg.jp domains.

3) Metadata & discoverability

- Cataloging exists (some portals implement DCAT-like catalogs), but not universally. Without stable dataset identifiers/versioning, reproducible analysis is hard.

4) Policy targets vs reality

- Digital Agency sets standardization and cloud migration targets (see policy pages). Progress is visible, but the operational reality—legacy on-prem municipal systems—means timelines and cost-savings are still a work in progress.

Technical examples (practical, copyable)

Fetch a time-series from e-Stat (replace YOUR_KEY):

curl -s "https://api.e-stat.go.jp/rest/3.0/app/json/getStatsData?appId=YOUR_KEY&lang=J&statsDataId=0003412312"

Quick Python pattern to load CSV and validate a schema (pandas + jsonschema):

from pandas import read_csv

from jsonschema import validate

url = 'https://example.gov/data/budget.csv'

df = read_csv(url)

schema = {

'type': 'object',

'properties': {'year': {'type': 'integer'}, 'amount': {'type': 'number'}},

'required': ['year', 'amount']

}

for _, row in df.iterrows():

validate(instance={'year': int(row['year']), 'amount': float(row['amount'])}, schema=schema)

要するに、このくらいの自動チェックを各自治体の公開フローに組み込めば、再利用は一気に楽になります。

Practical improvement roadmap

• Standardize dataset contracts: publish JSON Schema + CSV for common datasets (budgets, population, procurement).
• DCAT catalog + persistent dataset IDs across central and local portals.
• Reference open-source API gateway: one spec (OpenAPI) that municipalities can adopt; host a canonical mock server for developers.
• Git-based publishing + CI pipelines: data files + schema in repositories, automated validation, changelogs.
• Provide SDKs and examples: tiny Python/R/JS clients that talk to the common spec.

まとめ

これ見てくださいよ。国や自治体はダッシュボードで可視化を進めているけど、エンジニアが再利用しやすい形で公開するにはもう一段階の仕組みが必要です。APIの統一、機械可読なフォーマット、メタデータの整備、CIによる品質管理。技術で政策の実行力を上げられるポイントが山ほどあります。

おかむーから一言

テクノロジーで行政をアップデートするのは、単にツールを入れる話じゃない。API一本、スキーマ一つで市民の使いやすさが変わる。俺らエンジニアが一緒に作れば、もっと民主的で効率的な公共が作れますよ！