#2377: Accuratyping

You can now buy a typing tutor program which allows you to identify your individual error rate per finger, over a period of time.

The only data I can find indicate an error distribution of 2,4,1,2,0, 0,1,2,4,14 %.


I’m guessing this may vary significantly between individuals but it is an interesting measure of one’s physiology, which might in itself allow identification of an individual. Perhaps the motor homunculus in one’s motor cortex reflects these figures.

Today’s invention however is to gather the information for an individual who has not yet learned to type and then reprogram their keyboard so that the most frequently typed letters in one’s language, eg E, are placed close to one’s most accurate fingertip (and vice versa).

(There would be little point the indvidual whose map is shown above using his right, little finger to type E, for example).


  1. If you know these statistics per person you can improve the spellings correction.

    If a word is typed wrong and there are at least 2 fingers used and there exist correct words that could be typed with those same fingers, then the correct word with the HIGHEST finger error should be the most probable replacement.

    As keyboards cannot see the fingers this will be pretty difficult to implement. Therefore I propose the personal spelling checker in which the system holds lists of words you typed wrong and how you did improve them. This system learns the words you type even if they are not in a dictionary, it can automatically build up one, think of domain (job/company) specific words. This can even be done on sentence level with grammar checking, so the grammar checker learns your style of writing.

    • The highest finger error meaning the most frequent…yes, I get it. We seem to assume that all fingers are equally competent. This might be adapted for pianos and other dexterity-requiring operations. Your machine could also detect any neural deterioration shown by an increase in errors per finger.

Comments are closed