Castle Map: An Open-Data Atlas of 2,435 Global Fortifications

Exploring the World's Fortifications via Open Data

Castle Map is a free, interactive digital atlas that plots 2,435 of the world's most significant castles, fortresses, and palaces across 129 countries. The project serves as a curated visualization of global architectural history, allowing users to pan and zoom a night-mode map to access photos, founding dates, and historical narratives for each landmark.

Dataset Composition and Curation

The atlas is built entirely from open-source data to ensure transparency and accessibility. The specific data pipeline includes:

• Wikidata: Provides the core facts, founding dates, and geographic coordinates.
• Wikimedia Commons: Supplies the photographs for each entry.
• Wikipedia: Provides the historical stories and narratives.
• OpenFreeMap and Natural Earth: Provide the night-themed basemap.

To maintain an explorable user experience, the map is curated rather than exhaustive. To be eligible for inclusion, a landmark must have a verified photo, exact coordinates, and an English Wikipedia article. The current dataset consists of 937 castles, 389 fortresses, 692 palaces, and 417 ruins.

Global Distribution and Fame Ranking

France leads the atlas in total landmarks with 183, followed by Italy (179), Germany (176), and England (152). The project also includes significant sites from Japan (101), India (80), and Poland (77).

To rank the "fame" of these structures, Castle Map uses a proxy based on the number of Wikipedia language editions and sister projects that cover a specific site. According to this metric, the Palace of Versailles in France is ranked as the most famous castle in the world.

Technical Accessibility and AI Integration

Beyond the web interface, the project provides the full curated dataset as open GeoJSON and CSV files (CC0), which are mirrored on GitHub and Hugging Face. Additionally, the project offers an MCP (Model Context Protocol) endpoint, specifically designed to allow AI agents to programmatically access the castle data.

Community Feedback and Data Limitations

Following its presentation on Hacker News, the project received significant technical critique regarding its data coverage and categorization.

Data Omissions and Language Bias

Users pointed out that the reliance on English Wikipedia articles creates a systemic bias, omitting thousands of significant sites that are only documented in local languages. For example, users noted substantial omissions in Italy, France, and Ireland. One user highlighted that Italy may have up to 45,000 buildings that could fit the criteria, but many are missing because they lack English-language documentation.

Categorization Accuracy

Critics argued that the dataset occasionally miscategorizes landmarks or includes non-defensive structures. Specific examples cited by the community include:

• The Palácio da Bolsa in Porto: Identified as a 19th-century stock exchange rather than a castle.
• Hama-rikyū Gardens in Tokyo: Noted as a garden where the palace was destroyed in 1724.
• The North Carolina State Capitol: Questioned as to why it is categorized as a castle.

Developer Response

The developer, @Flightmussy, acknowledged that the site was "vibe coded" and initially developed with the assumption of low market demand. In response to the community feedback, the developer committed to improving the dataset by incorporating non-English Wikipedia sources and refining the categorization process, noting that the project is in its early stages and intended to make history more engaging.