Module 1: Introduction to GIS

1.2 A Brief History of GIS

From John Snow's cholera map to cloud-native tiles — how GIS evolved over 170 years.

Lesson 2 of 100·22 min read

Key takeaways

Spatial analysis predates computers by a century; John Snow's 1854 cholera map is a canonical example.

Modern GIS was born in the 1960s with the Canada Geographic Information System (CGIS).

The field has gone through four broad eras: mainframe, desktop, web, and cloud-native.

Introduction

To use any powerful technology well, it helps to understand where it came from. GIS is no different. The design decisions baked into today's formats, symbology choices, and workflows often trace back to specific people, institutions, and hardware constraints that existed decades ago. A short historical tour makes the modern tooling feel far less arbitrary.

We'll cover the pre-computer era, the birth of computational GIS, the desktop revolution, the internet and web mapping, and the current cloud-native, AI-influenced phase.

Before computers: 1854–1960

John Snow's cholera map (1854, London). During an outbreak of cholera in Soho, the physician John Snow plotted the addresses of deaths on a street map. Clusters of dots revealed that most cases clustered around a water pump on Broad Street. Removing the pump handle stopped the outbreak. Snow did not have a GIS, but his workflow — plotting georeferenced observations, looking for spatial clusters, forming a causal hypothesis — is the template of spatial epidemiology a century and a half later.

Sanborn fire insurance maps (1867–). Insurance companies needed to assess risk across US cities. Sanborn Company crews walked every city block, producing detailed street-level maps showing building materials, occupancy, and hazards. These are an early example of structured, updateable spatial data collection at scale.

Overlay on transparent acetate (1930s–60s). Landscape architect Ian McHarg popularised a technique in his 1969 book Design with Nature: draw each environmental factor (slope, vegetation, soils, flood risk) on a separate transparent sheet, then stack them to see where constraints overlapped. This is the direct conceptual ancestor of the raster overlay operations in every modern GIS.

The birth of computational GIS: 1960–1980

Canada Geographic Information System (CGIS), 1963. Roger Tomlinson — now called the "father of GIS" — led the development of CGIS for the Canadian federal government to inventory land use across Canada. CGIS introduced the vector data model, topological encoding, and the separation of geometry and attributes into linked tables. Many of its architectural decisions persist today.

Harvard Lab for Computer Graphics and Spatial Analysis, 1965. Directed by Howard Fisher, the Harvard Lab produced SYMAP (1967), GRID (raster analysis), and ODYSSEY (vector topological GIS). Harvard trained a generation of GIS thinkers including Dana Tomlin, Scott Morehouse, and Jack Dangermond.

Founding of ESRI, 1969. Jack and Laura Dangermond started the Environmental Systems Research Institute as a consultancy. By the 1980s it became the dominant commercial GIS vendor, releasing ArcInfo (1982) on UNIX workstations.

These early systems ran on mainframes, cost hundreds of thousands of dollars, and required specialist operators. A GIS department might have three people and one terminal.

The desktop revolution: 1980–2000

The personal computer brought GIS within reach of individual analysts. Key milestones:

ArcInfo (1982) — full-featured GIS on UNIX workstations.
MapInfo (1986) — one of the first desktop GIS products, popular with telecoms and insurance.
ArcView (1991) — simplified, map-centric Windows desktop GIS. ArcView made GIS mainstream inside municipal planning and environmental offices.
GRASS GIS (1982, open-sourced 1999) — originally developed by the US Army Construction Engineering Research Laboratory, GRASS became one of the first free, open-source GIS packages.
PostGIS (2001) — Refractions Research added spatial types to PostgreSQL, enabling enterprise-grade spatial databases on commodity hardware.

Two developments from this era still shape the field: the shapefile format (ESRI, 1993) became a de facto standard despite its many limitations, and Tomlin's Map Algebra framework (1990) gave raster analysis a rigorous vocabulary.

The web era: 2000–2015

The browser put maps on every desk. Milestones:

MapQuest (1996) and Yahoo Maps — early consumer web mapping, mostly server-rendered raster images.
Google Maps (2005) — the breakthrough moment. Slippy maps with 256×256 pixel tiles, smooth panning, and an accessible JavaScript API democratised interactive maps.
OpenStreetMap (2004–) — a Wikipedia-like collaboratively edited world map, now the primary open dataset for global street networks.
Leaflet (2011) — a lightweight, permissive-licensed JavaScript mapping library that became the default for open-source web maps.
MapBox (2010) — pioneered vector tiles, stylesheet-driven cartography, and developer-friendly APIs.

Simultaneously the OGC standardised server protocols — WMS (Web Map Service, 1999), WFS (Web Feature Service, 2002), WMTS (tiled maps, 2010) — enabling interoperability across vendors.

The cloud-native, AI era: 2015–present

Current trends:

Cloud-native raster formats — COG (Cloud Optimised GeoTIFF, 2018) and Zarr let applications stream only the pixels they need directly from object storage.
Cloud-native vector tiles — PMTiles (2022) packages an entire tile pyramid into one file accessible over HTTP range requests, removing the need for a tile server.
Vector tiles over raster — MapLibre GL, the open-source fork of Mapbox GL, renders vector data client-side at 60 fps with runtime styling.
AI and deep learning — segmentation of satellite imagery, automatic feature extraction, and conversational GIS assistants now sit alongside classical analysis.
Analysis-ready data — services like Microsoft Planetary Computer and AWS Open Data provide pre-processed Sentinel-2 and Landsat data through STAC APIs.
Browser-based GIS — tools like Atlas run full analytical workflows in the browser with no installs, collapsing the "desktop vs web" split.

A compressed timeline

Year	Event
1854	John Snow maps London cholera deaths
1963	CGIS — first computational GIS
1967	SYMAP — first widely used mapping software
1969	ESRI founded
1982	ArcInfo released on UNIX
1984	First GPS satellites
1991	ArcView brings GIS to Windows desktops
1993	ESRI shapefile format
1999	OGC WMS standard
2001	PostGIS 0.1
2004	OpenStreetMap launched
2005	Google Maps
2010	Mapbox introduces vector tiles
2011	Leaflet released
2014	MapLibre GL / Mapbox GL vector rendering
2018	COG (Cloud Optimised GeoTIFF)
2022	PMTiles single-file tile archives

Why history matters

Three practical reasons.

Format baggage. The shapefile's 10-character field name limit comes from a 1980s DBF constraint; GeoJSON's WGS84-only CRS comes from a deliberate 2008 decision to simplify the web. Knowing why helps you choose formats rather than inheriting defaults.
Methodological depth. McHarg's transparent overlays, Tomlin's map algebra, and Tomlinson's topology are still the conceptual core of the field. Tools come and go; these ideas stay.
Community. GIS is a relatively small, tightly networked field. Knowing who built what — Roger Tomlinson, Jack Dangermond, Mike Goodchild, Dana Tomlin, Helen Couclelis — helps you navigate conferences, literature, and communities.

Self-check exercises

1. What was John Snow's contribution to spatial analysis?

He plotted cholera deaths on a London street map in 1854 and identified spatial clustering around the Broad Street water pump, demonstrating that systematic spatial visualisation can expose causal hypotheses invisible in aggregate statistics.

2. Why is CGIS (1963) considered the first "real" GIS?

It was the first computational system to implement a vector data model, topological encoding, and the separation of geometry from attributes at national scale. It did what modern GIS still do: capture, manage, analyse, and display spatial data programmatically.

3. Name two cloud-native spatial data formats and what problem they solve.

COG (Cloud Optimised GeoTIFF) — lets clients read specific byte ranges of a raster file from object storage, so you only download the pixels you need. PMTiles — packages a whole vector or raster tile pyramid into one file accessed over HTTP range requests, removing the need for a tile server.

Summary

Spatial analysis is older than GIS software; John Snow and Ian McHarg established techniques still in use.
CGIS (1963) and the Harvard Lab founded the computational era.
The 1980s and 1990s brought desktop GIS and consumer-level accessibility.
Google Maps, OpenStreetMap, and OGC standards defined the web era.
Today's cloud-native formats, vector tiles, and AI workflows continue the trajectory.