Explore 5,793 castles, fortresses, palaces, and ruins through geography, founding eras, and the uneven trail of modern online attention.
Start here
A quick guide to this quest.
A castle can dominate a landscape yet barely register online, while another appears in dozens of Wikipedia language editions. This project connects the dataset’s physical coordinates and founding years with two measures of digital attention, while keeping missing and approximate dates visible.
Begin with the atlas
Circle size shows how many source records appear in each country. Select a circle to open that country’s highest-ranked landmark.
Climb the time tower
Compare broad founding eras, filter the discovery shelf, and see how the four landmark categories differ.
Enter the signal room
Compare Wikipedia language editions with recent pageviews, then save up to three landmarks in your personal trail.
Language reach
The number of Wikipedia language editions that have an article about a landmark.
Pageviews
Visits to a landmark’s Wikipedia articles during the trailing 365-day period captured by the source.
Fame rank
The source’s blended ordering based on language reach and recent pageviews; it is not a measure of historical value.
Founding year
The year attached to the landmark in Wikidata. Some dates are approximate and many records have no year.
Side Quest / VisualizationOpen the atlas, select a country marker, then switch to the Signal room to compare that landmark’s online reach.
5,793 places · 138 countriesWikidata snapshot · 31 July 2026
An atlas of old stone and present attention
Stone & Signal
Castles were built to control ground. Their digital afterlives compete for something else: attention. Explore where these landmarks stand, when they began, and which ones travel furthest online.
Landmarks in source5,793All four categories
Founding year listed3,6102183 have no year
Middle listed year1500Median of known years
Interactive set887of 887 curated records
01 / Atlas
Where the source records cluster
Each circle marks one country. Its area grows with the number of landmarks in the complete dataset. Select a circle to open that country’s highest-ranked place.
This coordinate plot is intentionally schematic: it shows latitude and longitude without national borders. Counts reflect the source, not a complete inventory of every surviving structure.
Discovery shelf
Across all eras
887 matches in the interactive set. Results are ordered by the source’s blended fame rank.
Why this exists
The question.
Historic landmarks occupy two landscapes at once: a physical place and a modern attention network. Putting location, age, Wikipedia language reach, and pageviews into one explorer can reveal where those landscapes align—and where they do not.
How to begin.
Open the atlas, select a country marker, then switch to the Signal room to compare that landmark’s online reach.
Method and limits
How the project was built.
I downloaded the 5,793-record World Castles snapshot linked by TidyTuesday and calculated country, category, and founding-era summaries from the full file. The interactive landmark set keeps the top 300 fame ranks, the 100 earliest dated records, 100 high-language/low-view records, every seventeenth fame rank, and at least two records per country; duplicates are removed, leaving 887 places. The map uses exact full-dataset country counts but schematic median coordinates. Missing years and pageviews remain missing, approximate years stay labeled, and logarithmic axes keep lower-attention records visible. No images or live requests are used.
What this demonstrates
Data analysis
Data engineering
Visualization
Information design
Interaction design
Editorial storytelling
Accessibility
Performance
Takeaways
Of 5,793 source records across 138 countries, 3,610 include a founding year. The median known year is 1500; the interface leaves the other 2,183 records in an explicit ‘Year unknown’ group.
The median listed founding year differs by category: 1293.5 for ruins, 1350 for castles, 1667 for fortresses, and 1716 for palaces. These are patterns in the curated source, not a timeline of all surviving structures.
Palaces have a median of 7,610 trailing-year pageviews in the source, compared with 2,586 for fortresses, 1,468 for ruins, and 1,169 for castles. Online visibility is not the same as historical importance.