At CARTO, they challenge themselves to use their platform as their users do, understanding better the pains and gains of using it. This approach helps them learn a lot about SQL, React, WebGL, projections, spatial algorithms, and mapping. They applied graph coloring to maps, which is a technique to assign colors to vertices in a graph so that no two adjacent vertices share the same color. Map coloring is an application of graph coloring, assigning different colors to adjacent polygons (countries, provinces, etc.). The Welsh-Powell algorithm is a greedy coloring approach, while Kempe's graph coloring algorithm adds nodes back to the graph in reverse order from which they were removed, coloring each added node with a color not used by its neighbors. These algorithms can be implemented as PostgreSQL functions using CARTO's batch SQL API. The resulting table contains a color assigned for each cartodb_id, and the original world_borders table can be joined with this table to visualize the result. The Kempe's graph coloring algorithm is more complex than the Welsh-Powell algorithm but still solvable using a pure PostgreSQL function.