It would prevent the common mistake of using QMatrix4x4 in a struct that is then copied into a buffer exposed as a uniform block in the shader. (unlike QVector2D and 3D which are 8 and 12 bytes, respectively, and thus are directly compatible with vec2 and vec3, QMatrix4x4 is has an extra flag, making it 68 bytes, which is unfortunate as it is incompatible with a mat4)