In the GIS world, for most modern formats and Standards (e.g., Shapefiles, WKB/WKT, GML, KML, etc.), valid polygons need to have closed linear rings, which is a linestring of coordinates where the first point is a repeat of the last point. For example, a triangle requires four points (not three).
POLYGON ((10 20, 30 60, 50 20, 10 20))
Who started this convention, and why? Is it some legacy of pre-Shapefile storage? (like how MS Windows still uses 2-byte CR+LF newlines?) Other non-GIS standards (e.g. SVG) don't require this repetition to encode polygons.