Street address errors in Google Maps

Nature of the Problem (Data vs “ML vibes”)

• Some argue the issue isn’t raw map data but an ML-driven “interpretation” layer that silently changes what users typed, similar to modern web search ignoring explicit queries.
• Others think the underlying data model is simply messy, with legacy design for a few Western countries stretched to global coverage and retrofitted with many edge cases.
• Several comments note that in Vancouver specifically, Maps clearly understands the numbering rules but appears to override them with bad “exceptions,” suggesting pipeline/heuristic issues rather than ignorance of the scheme.

Global Address Complexity

• Multiple commenters with professional experience (logistics, school districts, EMS, government GIS) stress that there is no single consistent notion of “how addresses work,” even within a single country.
• Examples: dual street names on county lines, roads with breaks, overlapping numbering on N/S segments, multiple valid city names per ZIP/postcode, buildings spanning streets, multiple historical or vanity addresses.
• International quirks: Japanese and Korean systems based on neighborhoods/blocks and permit order; rural regions with no street names or numbers; informal “directions by landmarks.”

Real‑World Failures and Safety Concerns

• Reports include:
  • Pins blocks or kilometers from actual buildings; address on wrong street segment or wrong city.
  • Driveways classified as public roads, parking lots mapped as through streets, stairs treated as drivable roads.
  • Destinations set to highway exits instead of train stations; malls/office buildings routed to taxi drop-offs instead of car parks.
  • Transit timing errors where layovers are misinterpreted.
• In some places (UK country lanes, Singapore, school districts, EMS), these mistakes are described as dangerous or operationally costly.

Quality Trends and Comparisons

• Some users claim Maps has significantly degraded in the last 2–4 years, calling it an “ML-brained mess”; others report years of flawless use.
• Apple Maps and OpenStreetMap are often cited as more accurate in specific locales (e.g., certain neighborhoods, new construction, some transit), though Google is praised for better POI search and some tricky formats (e.g., Queens, NY).

Feedback, Governance, and Abuse

• Many report edits being rejected, accepted but never applied, or oscillating between correct and incorrect states; fixes can take months or never land.
• There is concern that the same feedback mechanisms that let the author fix issues also allow malicious or careless edits; anecdote of a major road wrongly marked one-way causing citywide chaos.
• Some suggest Maps is now so ubiquitous it behaves like critical infrastructure and should be regulated, with mandated SLAs for corrections; others strongly oppose regulation.

Alternatives and Proposed Improvements

• Suggestions: better signaling when a result is a “guess,” route-aware search for stops along the way, distinguishing entrances (ride-share vs parking), more robust use of official GIS data, and leveraging user “home” locations to validate address placement.
• Alternatives mentioned include OpenStreetMap-based apps (e.g., OsmAnd), national codes like Eircode, and global schemes like Plus Codes/what3words, though adoption and UX remain challenges.