One thing these systems often overlook is that in the physical world symbols don't necessarily come aligned to invisible horizontal lines. For various reasons users may need to write vertically, along a curve, in a cross-word fashion...<p>A system where every symbol repeats with 90 degrees rotation multiple times is... not going to be very useful. Some writing systems like Arabic or Devanagari sidestep this by forcing the symbols to align on a line, but they sacrifice some functions to gain that.<p>Another aspect is kerning: when symbols don't uniformly fill the allocated space some pairs of symbols will inevitably look farther apart than others. Ideally, there should be as few such pairs as possible.<p>Of course, there are some other practical constraints like the direction in which the writer needs to move the writing tool, will the writer be able to see the symbols as they write them due to how they move the tool. How easy it is to confuse an essential element of a symbol for a decoration (what if you try to create a cursive version of a grotesque font? what if you want serifs?)<p>----<p>If I were to approach the task of creation of ideal writing system, I'd probably be looking at symmetries. An approach that would ensure that symbols are more or less unique under affine transformations. Another thing: something that eg. Japanese phonetic alphabet tried to do: group symbols by morphology. Eg. "B" and "P" should be more similar than, eg. "B" and "A", but there should also be a common theme for "B", "D", "G", "V" that's absent from "P", "T", "K", "F", and so on.<p>Also, there's really no reason to stick with Latin alphabet as the template. The I/J/Y, C/K/Q and V/U/W groups aren't good, just one character for the group is enough.