OpenStreetMap Explained

This article continues the Web Mapping & Tile Systems sub-hub with the foundational data source behind most of modern web mapping. Understanding OSM is essential for any developer building map-based applications.

The origin story

OpenStreetMap was founded by Steve Coast in 2004 in the UK. Coast was frustrated by the lack of free, openly-licensed street-level map data — in the UK at the time, the Ordnance Survey was the authoritative national mapping agency but its data was expensive to license. Even basic operations (driving directions, basic location services) were priced out of casual or hobbyist use.

Coast started OSM as “Wikipedia for maps”: a wiki-style project where anyone could contribute map data, and the resulting database would be free for anyone to use, study, modify, and redistribute.

Early contributors were:

• Cyclists mapping bike routes invisible on commercial maps.
• GPS hobbyists who had recorded tracks and wanted to share them.
• Map enthusiasts working from satellite imagery (initially Yahoo Aerial imagery, donated for OSM use).

By 2006, OSM had enough contributor base to establish the OpenStreetMap Foundation (OSMF), a not-for-profit organization registered in the UK, to provide legal and financial structure.

The growth

OSM's growth has been steady and substantial:

| Year | Registered users | Notable milestones | | ---- | ---------------- | ------------------ | | 2006 | ~10,000 | OSMF founded | | 2010 | ~250,000 | Reached 1 billion nodes | | 2012 | ~600,000 | License changed from CC-BY-SA to ODbL | | 2015 | ~2 million | Apple Maps starts using OSM substantially | | 2018 | ~4 million | Microsoft begins major contributions | | 2020 | ~6.5 million | Pandemic-era contributor growth | | 2025 | ~10 million | Continued growth |

Recent contributor counts:

• ~10 million registered users.
• ~2 million with at least one edit.
• ~100,000 monthly active contributors.
• ~10,000 daily active contributors.

Geographic distribution: contributors are heavily concentrated in Western Europe, North America, and Australia, with significant numbers in Brazil, Russia, India, Japan, and South Africa. Some regions have sparse contributor representation, leading to data quality variation.

The three-primitive data model

OSM's data model is remarkably simple — just three primitives:

Nodes

A node is a point with a latitude and longitude. Plus optional tags.

Examples of nodes:

• A specific tree (natural=tree).
• A stop sign at an intersection (highway=stop).
• A street-level address (addr:housenumber=123, addr:street=Main).
• A POI like a coffee shop (amenity=cafe, name=Starbucks).

A node has an ID (unique 64-bit integer), latitude, longitude, version (incremented on edit), timestamp, and the editing user.

Ways

A way is an ordered list of nodes. Ways represent:

• Lines — open ways like roads, rivers, fences.
• Polygons — closed ways (first and last node the same) like building outlines, lakes, fields.

A way is identified by its sequence of node references plus its tags (e.g., highway=residential, name=Main Street, maxspeed=30, surface=asphalt).

Most map features are ways: every road, river, building, and patch of grass on the map is a way.

Relations

A relation groups nodes, ways, and other relations into a larger logical structure. Each member of a relation has a role (e.g., “outer”, “inner”, “forward”, “backward”).

Common uses of relations:

• Multipolygons: a polygon with holes (e.g., a lake with islands). The outer ring is one way; the inner rings (holes) are other ways; the relation groups them with roles “outer” and “inner”.
• Route relations: a hiking trail or bus route with many constituent way segments. type=route, route=hiking, name=Pacific Crest Trail.
• Administrative boundaries: a country, state, or city boundary composed of many way segments.
• Restrictions: turn restrictions at intersections.
• Building groups: complex buildings with multiple components.

Relations let OSM represent complex spatial structures that don't fit a simple nodes-and-ways model.

Tagging

Tags are the key-value attributes attached to any element (node, way, or relation). Tags carry the semantic content: what is this feature, what are its properties.

Examples

highway=residential
name=Main Street
maxspeed=30
surface=asphalt
oneway=yes
lanes=2
sidewalk=both
cycleway=lane

building=yes
addr:housenumber=123
addr:street=Main Street
addr:city=Anytown
addr:postcode=12345
height=12
levels=3

amenity=cafe
name=Café Coffee
opening_hours=Mo-Fr 07:00-19:00; Sa 08:00-17:00
website=https://example.com
wheelchair=yes

The “any tags you like” philosophy

OSM's tagging system is deliberately open: contributors can use any key-value pair they want. This flexibility enables capturing unusual or new features without waiting for schema approval.

But the system would collapse into chaos without de-facto conventions. The community develops and documents conventions on the OSM Wiki:

• TagInfo (taginfo.openstreetmap.org) shows which tags are actually used and how often.
• OSM Wiki Map Features lists the standard tag vocabulary.
• Tagging mailing list discusses new conventions.

In practice, the top ~100 keys account for the vast majority of features. Following conventions is strongly encouraged for new contributors; deviating is allowed but reduces visibility (rendering tools target standard tags).

Licensing: ODbL

OSM data has been licensed under the Open Database License (ODbL) version 1.0 since September 12, 2012.

Before ODbL: CC-BY-SA

Originally OSM used Creative Commons Attribution-ShareAlike (CC-BY-SA). The change to ODbL reflected OSM-specific concerns:

• CC-BY-SA didn't clearly distinguish between database rights and content rights.
• Some uses of OSM data (combining with proprietary data) had ambiguous implications under CC-BY-SA.
• ODbL was specifically designed for databases.

The 2012 license cutover was contentious. Every contributor was asked to re-license their previous contributions; some refused. The data those contributors had submitted was removed from OSM when the cutover happened — leaving some regions with reduced coverage that took years to refill.

ODbL requirements

• Attribution: derivative uses must credit “© OpenStreetMap contributors” (and link to opendatacommons.org/licenses/odbl/ for the full license text).
• Share-alike: derivative databases must be licensed under ODbL.
• Database vs Produced Work: a rendered map is a “produced work” (not a derivative database) and doesn't need to be ODbL-licensed. You can use OSM data to make a proprietary rendered map.

Practical implications

• ✅ Use OSM data in any application (commercial or not).
• ✅ Render maps from OSM data and sell them.
• ✅ Build geocoders, routers, services using OSM.
• ⚠️ Distribute a database substantially derived from OSM: must license under ODbL.
• ⚠️ Use OSM with other databases and merge them: the merged result is ODbL.
• ❌ Strip attribution.
• ❌ Use OSM data without complying with the share- alike for combined-database use cases.

The license has been tested in court and is now well- understood in commercial OSM usage.

The Planet file

The Planet file is the full OSM database export, published continuously.

• Format: PBF (Protocol Buffers Binary Format) or XML.
• Size: ~150 GB compressed (PBF), ~2 TB uncompressed.
• Update frequency: full Planet weekly; minutely diffs for incremental updates.
• Distribution: planet.openstreetmap.org and many mirrors (Geofabrik, OSGF mirrors, university mirrors).

The Planet file is the canonical OSM dataset. Production OSM users typically:

1. Download a recent Planet snapshot or a regional extract.
2. Import into PostgreSQL/PostGIS using osm2pgsql or imposm.
3. Apply minutely diffs to keep the database current.
4. Render tiles, run geocoders, etc., from the PostGIS database.

Regional extracts

For applications that don't need the full Planet, Geofabrik provides daily-updated per-country and per-region extracts:

• Country extracts: e.g., Germany ~5 GB, US ~12 GB.
• State / province extracts: smaller files for US states, German Länder, etc.
• Continental extracts: aggregated regional data.

BBBike provides custom city-area extracts: specify a bounding box and download just that area.

Regional extracts have the same ODbL license as the Planet file. The smaller size makes them much more practical for development and small-scale production use.

Editing tools

OSM is edited through several tools, each with different audiences:

iD

The default web editor, accessible from “Edit” on openstreetmap.org. Beginner- friendly: drag-and-drop point/line/polygon creation, tag selection from predefined lists, photo overlay options.

iD is the most common editor for new contributors and accounts for the majority of all OSM edits.

JOSM

The Java Open Street Map editor: powerful desktop application for serious mapping. Features:

• Multi-window editing.
• Imagery layers (Bing aerial, Strava heatmap, custom WMS).
• Validation rules.
• Plugin ecosystem (hundreds of plugins for specific tasks).
• Conflict resolution for collaborative editing.

JOSM is preferred by power users and is the standard for complex mapping work (large-area imports, multi-day mapping projects, intricate multipolygon editing).

Vespucci

The standard Android editor. Field-mapping focus — record traces, take photos, edit while walking.

Go Map!! / OsmAnd

iOS editors for field mapping.

Rapid

A specialized editor with machine-learning suggestions (developed by Meta/Facebook). Suggests roads and buildings from satellite imagery; the human contributor verifies before committing.

StreetComplete

An Android app that asks bite-sized questions about specific features (“What is the street's maximum speed?”, “Is this building residential or commercial?”). Gamified, low-effort contribution. A major contributor to detailed-tag coverage.

Corporate contributors

Several major companies are substantial OSM contributors:

Apple

Apple Maps uses OSM as a foundational data source. Apple contributes significantly through paid mappers and through routine contributions by its mapping team. Apple is one of the largest corporate contributors by edit volume.

Amazon

Amazon Logistics contributes road and address data through paid mapping teams. Used to support Amazon delivery operations and AWS Location Service.

Meta (Facebook)

Rapid (the ML-assisted editor) is a Meta product. Meta contributes road and building data, particularly in developing regions, to support its services.

Microsoft

Bing Maps uses OSM. Microsoft contributes through paid mapping teams and through its satellite-imagery donations to OSM.

TomTom

TomTom is a long-time contributor and licensor of road data.

Smaller companies

Many smaller companies contribute: navigation startups, delivery services, real-estate platforms, ride-share companies.

Downstream consumers

The list of OSM consumers is extensive. Major examples:

• Apple Maps (foundational source).
• Microsoft Bing Maps.
• Amazon Location Service.
• Tesla in-car navigation.
• Snapchat.
• Foursquare / Swarm.
• Strava (athletic activity overlay).
• Komoot (hiking and cycling).
• Pokemon Go (Niantic Maps platform).
• Wikipedia / Wikimedia Maps.
• DuckDuckGo Maps.
• OsmAnd (open-source navigation app).
• Magic Earth (free navigation app).
• Maps.me / Organic Maps (open-source navigation).
• CyclOSM (cycling map).
• OpenTopoMap (topographic rendering).
• Many smaller services.

The OSMF maintains a list of corporate users at wiki.openstreetmap.org/wiki/Corporations.

Data quality

OSM data quality varies significantly by region:

High-quality regions

• Western Europe: comprehensive, frequently edited, often beats commercial data for footpath/ cycling detail.
• North America: strong road network coverage, good urban detail.
• Australia, New Zealand: comprehensive.
• Japan: detailed urban areas.

Moderate quality

• Eastern Europe, Latin America: good road networks, less complete detailed-feature coverage.
• Russia: variable; large cities well-mapped, rural areas sparser.
• Major Asian cities: well-mapped urban cores.

Lower quality

• Sub-Saharan Africa (variable): some regions well-mapped, others sparse.
• Some developing regions: less complete.
• Sparsely-populated areas: less contributor attention.

Quality assurance

The OSM community uses several QA tools:

• Osmose: automated error detection.
• KeepRight: validation rules.
• OSM Inspector: visual error inspection.
• MapRoulette: gamified fix-tasks.

Edit history is versioned (each element has a version number); any edit can be reverted. Vandalism is detected and reverted by community members.

Notable derived services

Nominatim

OSM's open-source geocoder. Powers OSM's own search, used by Wikipedia, and embedded in many applications. See /learn/what-is-geocoding.

Overpass API / Overpass Turbo

A query interface for OSM data. Use overpass-turbo.eu to write SQL-like queries against the live OSM database.

OSRM, Valhalla, GraphHopper

Open-source routing engines built on OSM data. Used in many navigation services.

Mapnik

The renderer behind OSM's default tile service.

OpenMapTiles, Protomaps

Vector tile generation projects (see /learn/vector-tiles-explained).

OSMF and the conference

The OpenStreetMap Foundation is the legal entity behind OSM. Key activities:

• Operating planet.openstreetmap.org.
• Hosting tile.openstreetmap.org (the public tile service).
• Maintaining the legal license framework.
• Sponsoring the annual State of the Map conference.
• Running working groups (Communication, Data, Engineering, Legal, License, Membership, Operations).

The State of the Map conference is the annual community gathering. Held in different cities each year (often Europe; sometimes North America, South America, Asia). Talks cover everything from data quality to corporate use cases to international mapping efforts. Many regional “State of the Map” events also occur (US, Europe, Africa, Asia, Latin America).

Common misconceptions

“OSM data is unreliable because it's crowd-sourced.” Quality is variable but often excellent. In many regions OSM is more detailed and current than commercial alternatives — particularly for footpaths, cycling infrastructure, and dynamic local features. The Wikipedia comparison is apt: imperfect but generally good and continuously improving.

“Big companies just take from OSM without giving back.” Not true for the major users. Apple, Microsoft, Amazon, Meta all substantially contribute via paid mapping teams and satellite-imagery donations. The OSMF Corporate Members program tracks contributing companies.

“OSM is anti-commercial.” No — ODbL explicitly permits commercial use. Many commercial services depend on OSM. The license ensures the data stays free; it doesn't prevent making money.

“OSM is just a hobbyist project.” With Apple, Microsoft, and Amazon as substantial contributors and consumers, OSM is far beyond a hobbyist project. It's a globally significant data infrastructure used by major commercial services serving hundreds of millions of users.

“OSM data is in Web Mercator.” No — OSM data is stored in WGS 84 latitude/longitude (EPSG:4326). Web Mercator is a rendering choice applied when generating tiles.

“You need permission to use OSM.” No — ODbL grants permission to anyone. Attribution is required; commercial use is permitted. No registration or approval needed.

“OSM tiles at tile.openstreetmap.org are unlimited.” No — the OSMF's tile service has a usage policy limiting it to small-scale use. Heavy users should self-host or use commercial providers. See /learn/tile-servers-and-caching-explained.

“Edits to OSM appear instantly.” Most edits are visible within minutes in the live database. The default tile rendering may take longer (15 minutes to several hours) for popular areas; rare areas may take longer. The data is current; the rendered tiles are cached.

“OSM only has roads.” OSM covers roads, buildings, businesses (POIs), boundaries (administrative, postal), natural features (water, forests, mountains), historic sites, addresses, trails, cycling infrastructure, transit routes, indoor mapping in some places, 3D building heights, and much more.

“The Planet file is too big to use.” For small applications: use regional extracts from Geofabrik (much smaller). For development: country extracts are typically 1-10 GB. State / province extracts are smaller. The full Planet is for global services.

“Apple just uses OSM directly.” Apple Maps uses OSM as one source among many, with substantial proprietary processing, refinement, and additional data layers (Apple satellite imagery, Apple-collected street imagery, third-party POI data). Apple's rendered maps don't look like raw OSM; the underlying data influence is present but not exclusive.

“OSM is a US project.” No — OSM was founded in the UK and the OSMF is registered in the UK. The contributor base is global, with Germany historically the largest contributor country. The project is genuinely international.

“OSM contributors all use the same editor.” Different tools for different audiences: iD for beginners, JOSM for power users, Vespucci and Go Map!! for mobile field mapping, StreetComplete for gamified contribution, Rapid for ML-assisted mapping. The editor ecosystem is diverse.