Extending that, heads have a much easier time moving left and right than up and down, since the motion uses the pivot joints. So, that means rastering left to right, then top to bottom, is the best match to the average reading and writing human (since right handedness is the dominant genetic trait [1]).

[1] https://www.nature.com/articles/s41598-018-37423-8

Top to bottom advantages everyone. Left to right advantages the right-handed. Right handed being the majority, top to bottom and left to right wins in almost every writing system.

And why would that make the top better than the bottom anyway? That's like saying the meal is worse after you finish it.

I don't think "natural" is used metaphorically. If you had an accurate simulation of the human hand you could show that one of the directions minimizes energy usage and damage to your hand, and I think it's the one we use. Starting high means gravity is helping you move down the page, and it's also easier to move your hand towards you than away from you, and the many small movements (rather than the one big one to the top of the next page) are where more energy is spent because of friction.

Writing is done by people and people are almost always subject to gravity. It's one of the 4 fundamental forces. Energy minimization is not an arbitrary selection criteria, it's central to the fitness/design of all living things.

I don't think it takes knowledge of gravity/energy/entropy to generalize that things more naturally "fall down" rather things naturally "rise up". But, it's probably a stretch to say that influenced writing direction.

Others have made a possibly more relevant point - in one direction, your arm/hand will block what you have already written.

yes and we daily see plants growing upward rapidly like 9.8 m/s²... maybe vapour and smoke going up are which we experience collectively as upward going things, but those are quite rare compared to like everything which falls to the ground.

yeah it's remarkable how many comments in this thread seem to be grasping onto random facts as if they represent a non-arbitrary justification.

is this a contrarian impulse or an anti-contrarian impulse?