It should be stated somewhere here that relativity operates in Minkowski space and not Euclidian 4D space.
What this means is that the length of a relativistic four-vector [ct, x, y, z] is sqrt((ct)^2 - (x^2 + y^2 + z^2)).
It is not the same as the Euclidian magnitude, i.e. NOT sqrt((ct)^2 + x^2 + y^2 + z^2)
The article does mention this, although with slightly confusing wording, but it should be emphasized that Lorentz invariant quantities need to be computed with the Minkowski metric.