Code-Driven Manifesto: Assessing Japan’s Government Data and APIs

どうも〜おかむーです！ Hey — Okamoo here, bringing a coder's eye to government data.

• Governments are publishing more datasets and APIs, but machine-readability and consistency remain uneven.
• e-Gov API catalog exists, some CSVs are open (Digital Agency, Cabinet Office), yet many useful reports stay PDF-locked.
• Technical fixes (standard schemas, OpenAPI, JSON/CSV-first publishing) could unlock rapid civic innovation.

結論

Japan has the foundations for a data-driven public sector — e-Gov portals, an API catalog, and pockets of CSV publishing — but the ecosystem needs stronger engineering practices: consistent machine formats, documented APIs (OpenAPI/JSON Schema), and operational guarantees (rate limits, versioning, uptime). 要するに、ポリシーはあるけど「コードで語る」仕様が足りないんですよね。

Report: what's actually available and what's missing

Inventory from public sources

• e-Gov Digital Agency API portal and API catalog (https://api-catalog.e-gov.go.jp/info/ja/apicatalog/list) enumerate many administrative APIs — good start.
• Ministry of Internal Affairs and Communications (Soumu) pushes an open data strategy emphasizing common APIs and standard data models.
• Practical CSVs exist: Digital Agency hospital CSV, Cabinet Office economic time series CSVs (example: https://www5.cao.go.jp/j-j/wp/wp-je24/csv/d1-1-4.csv), MOFA country-list CSVs. These are gold for devs.

これ見てくださいよ：一方でPDFでしか公開されてない報告書や、CSVがあるけど列名がバラバラ、日時フォーマットが統一されていないデータがまだ多いんです。

Machine-readability and formats

• Positive: CSVs on digital.go.jp are great — straightforward for Pandas.
• Negative: PDFs and inconsistent CSV schemas force manual cleaning. No central schema registry means each project re-invents parsing logic.

Example: quick Python snippet to fetch a public CSV and inspect columns

import pandas as pd
url = 'https://www.digital.go.jp/assets/.../xxxxxx_hospital.csv'
df = pd.read_csv(url)
print(df.columns)

API quality: docs, specs, and operational behavior

• Catalog exists, but not every API publishes an OpenAPI/Swagger spec or machine-consumable schema.
• Questions to ask: authentication model, rate limits, pagination, error codes, CORS — these must be standardized so frontend and backend teams can build reliably.

Data quality and policy targets

• Policy documents often specify targets (e.g., open data coverage, API rollout), but dashboards linking targets to time-series data are rare.
• Practical check: if you want to track progress, you need time-stamped, versioned datasets with provenance fields (published_at, source_url, dataset_version).

Concrete engineering recommendations

• Publish OpenAPI specs for every API in the e-Gov catalog. This enables codegen, client SDKs, and tests.
• Adopt a minimal common schema set (JSON Schema + CSV dialect): date/time in ISO8601, prefixed geocoding fields (lat/lon), standard address components.
• Prefer JSON/CSV-first publishing. When PDFs are necessary, attach machine-readable artifacts (CSV/JSON) alongside.
• Run CI for data quality: validate new CSVs against schemas, check for null rates, enforce column names.
• Provide a schema registry and catalog with versioning + dataset provenance metadata (DCAT or schema.org Dataset).

Example OpenAPI fragment (conceptual)

openapi: 3.0.1
info:
title: Prefecture Hospital API
version: '1.0.0'
paths:
/hospitals:
get:
parameters:
- in: query
name: prefecture_code
schema:
type: string
responses:
'200':
content:
application/json:
schema:
$ref: '#/components/schemas/HospitalList'

Use cases unlocked by improvements

• Real-time dashboards for disaster response using standardized location+capacity APIs.
• Faster civic apps: developers can scaffold frontends from OpenAPI clients instead of writing fragile scrapers.
• Better policymaking: link targets and actuals by ingesting time-series CSVs with clear provenance.

まとめ

Japan's public-data layer is promising: catalogs and CSVs exist, but we need engineering discipline to scale. Publish specs, standardize schemas, ensure machine-readability, and automate quality checks. 要するに、政策をコードで語れるようにするだけで、社会実装の速度と信頼性は格段に上がりますよ！

おかむーから一言

I've built startups and shipped gov-facing systems — tech fixes here are low-hanging fruit for huge impact. Let's make government data friendly to code and to people!