Flattening a sphere onto a map always distorts something — area, angle, or distance has to give

What it is

The Earth is a curved surface; a map is flat—flattening a sphere necessarily distorts something, a geometric inevitability rather than a shortcoming of cartographic craft. It is mathematically proven that no planar projection preserves area (equal-area), angles (conformal), and distances (equidistant) all at once; every projection trades among the three. Equal-area projections keep areas but distort shapes; conformal projections keep shapes and directions but inflate areas; equidistant projections are accurate only for specific points and lines. Choosing a projection means choosing what to sacrifice—and the sacrifice directly affects the fairness of the data visualization drawn on the map.

Why it happens

The inevitability has an intuitive form: the orange-peel problem—a sphere cannot be flattened without tearing, so something must stretch or compress somewhere. Rigorously, the Theorema Egregium shows a surface's Gaussian curvature is invariant under isometry; the sphere's curvature is positive and the plane's is zero, so no mapping can preserve all measurements. Projection distortion is not uniform: most projections distort differently by location (Mercator is near-accurate at the equator and wildly inflated at high latitudes), meaning the trustworthiness of area relationships varies across the same map. The impact on data visualization depends on the encoding: area encodings (choropleth maps filling regions by value) are most sensitive—if a projection inflates a region's area, the reader's sense of its value inflates too; position encodings (point markers) tolerate area distortion but suffer under distance distortion (distance and buffer analyses need equidistant or equal-area); direction and route analyses need conformal. So the first question of projection choice is not "which projection is best" but "which visual channel encodes the data on this map."

Where it stops holding

The trade-off's severity scales with coverage: for small-extent maps (a single city), any reasonable projection's distortion is negligible and projection choice barely matters; distortion becomes a design factor only at continental or global extents. Compromise projections (like Robinson—preserving no property but balancing all distortions) are designed for visual impression rather than mathematical properties, suited to general reference maps, not precise comparison. Spatial analysis (not just display) is affected too—algorithms computing area, distance, or density must run in a coordinate system appropriate to the data, or the analysis itself carries systematic bias.

Applying it

• Before choosing, answer: which channel encodes the data (area / position / direction)? Select the projection family by channel.
• Use equal-area families (equal-area cylindrical, Albers, Mollweide) for global or large-extent choropleths.
• Run spatial analysis (distance, area computation) in an equal-area or equidistant coordinate system; the display projection and analysis projection may differ.
• Verification: compare the same data's area impression under two projections; if readers' conclusions flip with the projection and the caption does not declare it, governance is missing.

Related

• Same group: U12.01.2 Area-type maps must use an equal-area projection · U12.01.3 The Mercator projection severely inflates areas at high latitudes · U12.01.4 Projection choice must change with the map's coverage · U12.01.5 The projection used should be stated in the caption
• Nearby: U12.01.2 Area-type maps must use an equal-area projection · U12.01.3 The Mercator projection severely inflates areas at high latitudes
• Search terms: map projection · equal-area · projection distortion