A Study on Typing

I'm a pretty proficient typer, but I've definitely noticed that some words are harder than others. Which begs the question, what makes some words hard to type, and what makes some words easy to type?

Intuitively, I think it's probably about the distribution of letters and more specifically where they are positioned on the keyboard. A naive answer might be something like: time to type is roughly proportional to the total distance between the letters.

A more nuanced hypothesis might be that what's more important is the alternation of hands, because that's what allows typist to start really flowing between the characters. Double letters are probably a single finger needs to take on so much responsibilty (try typing "balloon" really fast, you'll notice your ring finger is noticebly overworked).

To answer this question, we need to do some data collection first!

After doing this for about 30 minutes, I got some results that we can start to analyze.

From the chart, we can see that most letters took me about ~55ms to type, and almost all of them are below 200ms. As we might also expect, the frequencies also form a normal distribution with some outliers towards the right. (Note that in this graph, I have pruned datapoints that were more than 5 standard deviations above the mean which removed 139 datapoints)

Now considering the average time a pair of letters takes, we notice that some take much longer than others (looks like Zipf's distribution?). This begs the question, what causes some pairs to take more time than others?

My first instinct is to look at the distances between letters on the keyboard, perhaps there is some way to correlate how far apart letters are with the time it takes to type them. Obviously this is a pretty crude approximation since we type with two hands (hopefully) so we don't really care about the absolute distances very much but it's worth taking a look at.

I guess we can't reject the null hypothesis here. In fact there is a negative correlation between distance and time to type. May this could be explained by the fact that letters that are far apart tend to be typed by different hands and therefore are actually faster not slower.

To test this theory, we can count the number of times we have to switch hands when typing a word (normalized by length of word), and do the same analysis. If words that require more hand switching are quicker to type, that proves some supporting evidence my explaination.

Well, not really. As it turns out, it's more difficult than I thought to find correlations between how long it takes to type a word and simple metrics about the word. It'll take some more effort to draw interesting conclusions about what factors are important here.