In late 2025, Parks Canada quietly notified its provincial and territorial partners of an impending decision. The Canadian Register of Historic Places — historicplaces.ca — would be decommissioned in spring 2026. The register, launched in 2004, was the single national database of approximately 13,500 historic places formally recognized by federal, provincial, territorial, and municipal governments across all thirteen provinces and territories. It was the only resource of its kind: a comprehensive, searchable record of what Canadians at every level of government had chosen to formally designate as worth preserving.
Parks Canada cited technological obsolescence, security vulnerabilities, and outdated coding as the reason. The site, they said, had become too technically complex and costly to maintain.
Parks Canada announces the shutdown of historicplaces.ca#
On January 25, 2026, CBC Nova Scotia broke the story to the public. The reaction from the heritage community was immediate and unambiguous. The National Trust for Canada issued a formal alarm and wrote to federal, provincial, and territorial ministers calling on them to ensure the data remained accessible. Emma Lang, Executive Director of the Heritage Trust of Nova Scotia, called the database’s importance “incalculable,” noting that it was used by everyone from genealogists to Realtors to architects, engineers, and heritage professionals. Provincial heritage officials across the country described their initial response as shock and disappointment.
The scope of the loss was significant. The register contained detailed records for each designated site — not just names and addresses, but heritage value statements explaining why each place matters, character-defining elements describing the specific architectural and structural features worth preserving, historical and current function classifications, thematic categories, construction dates, recognition dates, architect and builder attributions, and multiple photographs. This information had been assembled over two decades through the painstaking work of heritage professionals at every level of government. Much of it existed nowhere else in a structured, searchable format.
Nova Scotia, which lacks its own provincial heritage website, stood to lose access entirely. Prince Edward Island began an emergency review of its own heritage database. Manitoba was better positioned only because the Manitoba Historical Society had independently maintained its own provincial database. Across the country, officials scrambled to understand what they were about to lose.
Parks Canada’s replacement: Excel spreadsheets without images#
Parks Canada’s proposed replacement was a narrower database covering only federally registered properties — a minority of the 13,500 listings. Federal designations represent the smallest slice of the register. The vast majority of designated heritage sites in Canada are recognized at the provincial, territorial, or municipal level. These would simply disappear from any national database.
For the provincial and territorial data, Parks Canada offered to provide Excel spreadsheets to participating jurisdictions. The spreadsheets would not include images. They would not include the heritage value statements or character-defining elements in a structured format. They would not be searchable by the public. Two decades of carefully assembled, publicly accessible heritage data would be reduced to rows in a spreadsheet sitting on a government file server somewhere, accessible to no one outside the heritage bureaucracy.
Rebuilding 13,554 heritage sites in 24 hours#
When I read the CBC article, I felt a significant urgency to act. This was a national heritage resource about to disappear — not because the data was lost, but because the government had decided that maintaining a website was too difficult. The data was still there. The images were still there. The heritage value statements were still there. Someone just needed to grab it before the plug was pulled.
That weekend, I built Heritage Guide Canada. Using Claude Code and a Python scraping pipeline, I downloaded the entire database — all 13,554 English site records, all 13,558 French site records, and all 55,069 images (both thumbnails and full-resolution versions) — in under six hours. I then built a complete replacement application and had it live within 24 hours of the CBC story’s publication.
The urgency was real. Parks Canada had set a timeline. Every day the original site remained up was a day the data could still be captured. I did not know exactly when the shutdown would happen, and I was not willing to find out by waiting.
Heritage Guide features: search, maps, and bilingual access#
Heritage Guide is a bilingual (English and French) web application that preserves the complete Canadian Register of Historic Places and improves upon it in several significant ways.
The search system offers multi-faceted filtering that the original government site never had. Users can simultaneously filter by province or territory, construction period (ranging from 1605 to 2023), recognition date (from 1905 to 2025), historic function (over 100 categories including railway stations, churches, lighthouses, forts, banks, and Aboriginal ritual sites), and thematic category (approximately 25 themes such as architecture and design, military and defence, trade and commerce, and communications and transportation). The original site offered basic keyword search that required consecutive, exact-match phrases — a search interface that had not meaningfully changed since the mid-2000s.
Beyond keyword search, Heritage Guide includes AI-powered semantic search. Users can search for concepts — “fur trade” or “Victorian architecture” or “wartime industry” — and the system returns relevant results based on meaning rather than exact keyword matches. This is powered by vector embeddings generated for every record in the database, with an in-memory approximate nearest neighbor index for real-time similarity search.
Each heritage site page presents the full data profile: all available photographs, the heritage value statement explaining why the place matters, the character-defining elements that define what should be preserved, recognition details and dates, architectural attributions, historic and current function classifications, thematic categories, and geographic coordinates. This is the information that Parks Canada’s Excel exports would have omitted or stripped of structure.
Heritage Guide adds several features that the original site never had. A Near Me tool uses browser geolocation to find designated heritage sites in your immediate area. An interactive Explore Map lets users discover sites geographically across the entire country, with client-side filtering by construction period, recognition date, function, and theme — all rendered simultaneously for the full 13,554-site dataset. Curated city pages aggregate heritage sites by municipality, making it easy to explore the designated heritage of any major Canadian city. Browsing by building type — railway stations, churches, lighthouses, forts, banks, theatres — makes the large dataset navigable by category.
The entire application is bilingual. Every page, every search query, every filter label, and every navigation element works in both English and French. Each heritage site has its own URL in both languages, with proper cross-language linking for search engine indexing.
Hosting a national heritage database on a Raspberry Pi#
The production deployment of Heritage Guide runs on a Raspberry Pi — a single-board computer roughly the size of a credit card that costs less than a restaurant dinner. The site serves over 27,500 prerendered static HTML pages (every heritage site in both languages, plus city pages, topic pages, and static pages) through nginx, with a lightweight Node.js API server handling search queries and dynamic operations. Images are served from an S3 bucket to keep bandwidth manageable on the Pi’s connection.
This is the detail that matters most in the broader context. Parks Canada told the heritage community that historicplaces.ca had become too technically complex and too costly to maintain. Heritage Guide demonstrates that a single developer, using modern tools, can build a better version of the site in a weekend and host it on hardware that draws less power than a light bulb. The technology to do this is not exotic. The barrier was not technical complexity. It was institutional will.
Why government digital infrastructure should not be disposable#
Government digital infrastructure does not have to be fragile, expensive, or disposable. The pattern of building government websites at enormous cost, allowing them to decay for a decade, and then decommissioning them because maintenance has become too difficult is not inevitable. It is a choice — a choice that reflects how governments budget for technology (large capital expenditures with no provision for ongoing maintenance) and how they value the digital public goods they create.
The Canadian Register of Historic Places was assembled over twenty years through the collaborative effort of heritage professionals at every level of government across the country. Each record represents a formal decision by a community, a province, or the nation that a particular place is worth recognizing and preserving. That collective work has real value — not just to heritage professionals, but to anyone who wants to understand the built history of where they live.
Letting that resource disappear because of a server migration is not a technology problem. It is a priorities problem. And the fact that one person with modern tools could rebuild the entire thing in 24 hours makes the original decision to decommission look not just unfortunate but unnecessary.
CBC, Radio-Canada, and national media coverage#
CBC featured Heritage Guide in a full article: “Parks Canada planned to delete a heritage website — but one proud Canadian hit ‘Save.’” Radio-Canada published two separate pieces — one covering the project directly and a follow-up exploring how AI is becoming an unexpected ally for Canadian heritage preservation. The story was syndicated by Yahoo News Canada and picked up by international outlets. On LinkedIn, the announcement generated significant engagement, including a comment from Brian Anthony, who was directly involved in the original Historic Places Initiative and praised the effort.
I have offered to give the site to the government if they find it useful. Heritage Guide is an independent initiative with no affiliation with Parks Canada or the Government of Canada. The data originates from official government sources and belongs to all Canadians. It should remain accessible regardless of what happens to any particular government website.
Community contributions and heritage data maintenance#
Heritage data is not static. Buildings get renovated, demolished, or repurposed. New sites get designated. Existing records contain errors that heritage professionals notice. Heritage Guide includes a contributor system that allows registered users to submit corrections and additions to the data — text edits, new photographs, and location corrections. Submissions enter a review queue with before-and-after diffs for admin approval. This creates a path for the heritage community to maintain and improve the data over time, rather than depending on a single institution to do so.
What’s next for Heritage Guide Canada#
Heritage Guide is live now at heritageguide.ca. The semantic search system is operational. The Explore Map covers all 13,554 designated sites across every province and territory. The bilingual interface serves both English and French Canadians. The contributor system is accepting registrations from heritage professionals.
I am continuing to develop the platform — refining the semantic search, adding curated topic articles about Canadian heritage themes, and working to ensure that the data remains accessible for as long as it is useful. The Canadian Register of Historic Places took twenty years to build. It should not take a weekend to lose.
For the technical details of how Heritage Guide was built — the Python scraping pipeline, the MongoDB architecture, the React SSR system, the AI-powered vector search, and the Raspberry Pi deployment — see the companion post: Building Heritage Guide: A Technical Deep Dive.
