It's an interesting approach, and I feel like it's onto something with the algorithm. Bit like the flashcard programs in that regard.
It might be worth generating words in a natural language structure. I don't know about others but I tend to read a little ahead of what I'm typing. Pick up the words rather than the letters. When the words are fairly rare, (lots of these words are, at least for me,) and don't seem to follow a linguistic structure, this reduces to reading letter by letter fairly quickly and then searching the string for the next letter.
Like, normally my brain would be going 'This is a thing and my fingers are just filling in in the background. They're not suffering from my conscious attention.'
And at the moment it's going 'o-r-g-o-t' ... 'e-u-g-e' ... 'g-e-r'
It might be worth generating words in a natural language structure. I don't know about others but I tend to read a little ahead of what I'm typing. Pick up the words rather than the letters. When the words are fairly rare, (lots of these words are, at least for me,) and don't seem to follow a linguistic structure, this reduces to reading letter by letter fairly quickly and then searching the string for the next letter.
Like, normally my brain would be going 'This is a thing and my fingers are just filling in in the background. They're not suffering from my conscious attention.'
And at the moment it's going 'o-r-g-o-t' ... 'e-u-g-e' ... 'g-e-r'