There Are No Clean Maps: Reconciling Incompatible Geographies

We often speak about “putting data on a map” as if geography were stable and agreed upon crisp boundaries, clear hierarchies, seamless layers. In practice, there are no clean maps. Every map is a negotiation between data sources, definitions, and vintages. Maps are not neutral mirrors of reality; they are reconciliations built on trade-offs and hard decisions.

This talk explores what it means to build spatial analysis when underlying geographies do not align across time, hierarchy, or definition.

Drawing from my work with village-level, pincode-level, hex-level, and district-level data, I unpack the analytical decisions required when:

1. Administrative hierarchies don’t align across time Village shapefiles from 2011 coexIst with district boundaries updated in 2022. The hierarchy is conceptually nested but spatially inconsistent. What assumptions are embedded when we visualise one inside the other? How do these mismatches affect interpretation?

2. Parallel geographies coexist. Districts and pincodes are both valid units, yet neither nests cleanly within the other. When users need insights at both levels, we are forced to construct relationships that do not naturally exist.

3. Granularity must be constructed Data rarely comes at the level we need. To make it usable, we’ve had to create new spatial units to cover gaps, estimating population at village and pincode levels, and combining multiple sources to produce consistent analytical layers.

4. Spatial logics conflict Datasets such as WorldPop operate on raster grids, while other datasets may rely on hex-based systems like Kontur’s hexes. Merging grids and hexes means translating between fundamentally different spatial assumptions about area, density, and aggregation.

5. Geographic provenance is unknown Survey data often arrives at the district level without clarity on which boundary definitions were used. When boundaries evolve, a “district” is not a stable object. How do we responsibly visualise such data when the shapefile itself is uncertain?

Across these cases, mapping is not just a visual exercise but a series of analytical trade-offs. It requires deliberate choices about boundaries, aggregation, approximation, and how much uncertainty we are willing to carry forward.

Rather than presenting a single solution, this session argues that spatial visualisation begins long before design; in the invisible work of reconciling geographies. The question is not whether trade-offs exist, but whether we acknowledge and communicate them.