Why maps lie about size
Every flat map of a round planet distorts something, and the most common ones distort size. The distortion is not a mistake, it is the price of keeping angles true, but almost nobody is told they are paying it.
A round planet on a flat sheet
You cannot flatten a sphere without stretching it. That is not an engineering limitation, it is a geometric fact, and every map projection ever drawn is a choice about where to put the stretching.
The projections used for most world maps preserve angles, which is enormously useful for navigation: a straight line on the map is a constant compass bearing at sea. The price is that scale grows with distance from the equator, and it grows fast. At sixty degrees of latitude the map is stretching everything by a factor of two in each direction. Nearer the poles it is far worse.
Nobody is hiding this. It is just that a map is usually the first thing a child sees about the shape of the world, and the caveat never arrives.
Greenland, the standard example
Greenland spans 2,651.5 km north to south. Africa spans 7,887 km north to south. On a classroom world map those two shapes look comparable, and the real ratio is very nearly three to one on that axis. By area the gap is far wider still.
The reason Greenland is the standard example is that it sits almost entirely above sixty degrees north, where the stretching is at its most aggressive, while Africa straddles the equator, where there is none. The two worst-case and best-case positions on the same sheet of paper.
Iceland is the same trick at a smaller scale. Iceland spans 512.2 km east to west and Greenland 2,651.5 km north to south. A ratio of 5.18, between two places that appear on the same maps at roughly the same latitude, so the projection treats them consistently and this particular comparison survives it.
Australia against everything
Australia is the counterexample that surprises people in the other direction. It spans 4,035.6 km east to west across the mainland, and on most world maps it looks like an island tucked into a corner rather than the continent-sized landmass it is.
Canada spans 5,395.3 km east to west, a ratio of 1.34 over Australia, which is far closer than the map suggests because Canada gets the northern stretch and Australia does not. Africa against Australia is 1.95 on these axes, and again the map exaggerates the gap in the wrong direction.
The most useful anchor of all is not a country. The Moon is 3,474.8 km across, so the east-west span of mainland Australia exceeds the diameter of the Moon by about sixteen per cent. Neither object appears on a world map, so the projection cannot interfere, and the comparison survives intact.
American states against European countries
This family of comparison is popular and mostly reliable, because the places involved sit at similar latitudes and get similar treatment from the projection.
Texas spans 1,244.8 km east to west and the United Kingdom 1,220.5 km north to south, a ratio of 1.02, which is as close as the catalogue gets outside deliberate near-ties. Italy at 1,288.9 km north to south against California at 1,060.8 km north to south is 1.22, and there both figures are on the same axis, which makes it the cleaner of the two.
The caveat is the axis, not the projection. Texas measured north to south gives a different number, and comparing a Texan width to a British length is arithmetic rather than geography. The page says which axis each one uses for exactly that reason.
Why span and not area
Area is the figure most people reach for, and it is a perfectly good one. This game uses span instead, for a reason that is about drawing rather than about geography.
The game asks you to resize an outline until it looks right. That is a one-dimensional judgement: you are dragging a handle. Scoring it on area would mean the score moved with the square of your error, which makes near-misses feel like disasters and makes the whole thing miserable. Scoring on a single span keeps the error linear and the feedback legible.
The consequence is that a long thin country and a compact one of the same area get very different figures, and that is a real limitation worth naming. Chile and Belgium cannot be usefully compared by span. Nothing about the method pretends otherwise.
What the outlines are built from
The 118 geographic shapes in the catalogue are built from open mapping data rather than drawn by hand, which matters more than it sounds. A drawn coastline is a stylisation, and a stylisation of a coastline is almost always smoother and rounder than the real thing, which changes the proportions of the shape.
Because the outlines come from mapping data, the shape you are judging against is the real one, and a comparison like Russia at 8,082.3 km east to west against the contiguous United States at 8,055.6 km can be trusted at the level of a per cent. That ratio, incidentally, is 1.00 to two decimal places, which is the sort of thing you only discover by measuring.
What to carry away
Two things. First, any size intuition you have about the world that came from a flat map is unreliable above about forty degrees of latitude, and badly unreliable above sixty. Second, the fix is not a better map, because there is no better map, only different compromises. The fix is a number with an axis attached to it.
That is unglamorous compared to an animated projection that morphs one country onto another, and it is the only version that survives being written down.