The easy way to mentally visualize it is to think that the stylus has "sweet spots" that has to be in contact with the groove for the sound to be close to optimal. Tilt the cartridge a bit, and the sweet spots on both sides are more or less there. But even the slight subsequent rotation around the vertical axis would move the point of contact between the stylus and groove wall away from our desired spot.
When we design or set up an arm, it is reasonably easy to minimize static geometric errors. However, we must always keep in mind how these small static errors will be compounded by dynamic ones.
Inner groove distortion with a typical 9" pivoted arm is a typical example of such interaction. 1 degree of alignment error is worsened by lower linear speed of stylus in groove. LP eccentricity is, proportionally to radius, much higher in inner grooves. It induces scrubbing and torsional cantilever motion -- and you can get total alignment error so significant that it might result in mistracking.
Textbook approach to solve this mistracking is to minimize angular error in inner grooves. However, this does nothing to address dynamic errors.
Sometimes, one can improve things by aligning and realigning the cartridge, using various geometries etc. But most often, you cannot get rid of inner groove mistracking by altering horizontal alignment. It isn't the only source of error, and changing it alone might not provide a solution.
A good tonearm is designed to satisfy two often mutually exclusive requirements: first, to provide acceptable static geometry and second, to minimize dynamic geometric errors as it follows the groove on the imperfect disc. This is the reason why many Japanese arms (SAEC is a good example) do not conform to established Stevenson, Löfgren etc geometry calculations. Their designers sacrificed some static precision to obtain better dynamic behaviour.