REINVENTING HOW WE TYPE

As humans, we have so many thoughts and ideas to share, and we have whole bodies to do so--it seems ridiculous that we use only our fingertips to express most of these thoughts into the world. We take our grand, abstract ideas and funnel them through the arduous, inefficient process of typing. Not only that, but we're using a QWERTY layout that was invented in 1874! Something feels wrong about that.

PART 1: examining the qwerty keyboard

There are a plethora of articles and opinions criticizing the QWERTY layout, mentioning how it favors the left hand, overuses our weaker fingers, forces ulnar deviation of the wrist, has a weird, rarely-used consonant hodgepodge on the home row, etc.

I looked up the DVORAK keyboard because I'd heard it solved many of QWERTY's problems, but turns out it hasn't changed since 1936! It's baffling how we're okay with that, but are upset that the iPhone 8 'isn't enough of an upgrade.'

Tech has changed and its input mechanisms should change too. We don't have to interact with language the same way we did 100 years ago. And that means questioning many basic assumptions about QWERTY, keyboards as a whole, and even text itself.

I wanted to start by looking at typing speed. But not overall typing speed, I specifically was curious about sequences of letters. I wanted to see which letter combos are the fastest to type, ergo most efficiently placed, and why? Conversely, what makes slower letter combinations so inefficient? Are they right next to each other or really far apart? Is there an ideal distance or placement?

My Qs:

• What are the fastest and slowest letter combos to type, i.e. which letters are most conveniently (and inconveniently) placed?
• What does this tell us about the rhythms of our hand/finger movements? (and how can this be used to design a better layout?)

I predicted that some key combos would be much faster to type than others, perhaps because they have an "ideal distance" between them on the keyboard—or perhaps because they use a natural finger rhythm, kind of like drumming your fingers. Try typing 'e-r-g' in a natural, falling finger movement--you'll notice 'i-n-g', which is infinitely more common, isn't nearly as easy to type.

 The rhythm argument is pretty huge, I think. As with boxing, or walking, or any number of body movements, typing has a cadence. Alternating hands/fingers when typing improves flow because when one finger is pressing a key, another one can be cocked into position for the next keystroke. Because of this, I expected that doubles (ss, ee, rr) would be one of the slowest things to type, since you have to lift the same finger up and bring it down again. A keyboard layout that is optimized for smooth finger rhythms would easily be king.

So I wrote a program in Python that takes takes in a bunch of real-time typing data. It then plots a histogram of the average time it took the user to type each letter pair, or bigram. Since we often use special keys, like 'Backspace' and 'Enter,' I wanted to include all of those in this analysis as well. ('space' is represented by _/ in the graphs below)

Here's an example of the output using a sample sentence: The quick brown fox jumped over the lazy dog

To get some more complete data, I then typed the first page of David Eagleman's new book, The Runaway Species

I looked closer at the data:

As expected, many of the slowest bigrams included some sort of special character, such as 'f' followed by a colon. Those are naturally slower to type since we use them less and they are placed off to the side--not to mention they sometimes involve the 'Shift' key as well.

But what was also interesting was that many of the slowest pairs involved keys on the edges of the bottom row, aka the ones that force us to awkwardly crimp our ring and pinky fingers: z, m, period, comma. Another combo I noticed was 'lu' -- certainly an awkward motion typing that one.

When I looked at the fastest bigrams to type, it was obvious that rhythm was a factor, since many of them were pretty smooth combos to type. They all either use an alternating hand motion (di, wn) or, if they use the same hand, are placed in such a way that they elicit a natural drumming motion (fr, th, hu).

And the medium speed bigrams were just that -- not too awkward, not too smooth.

Comparing across the spectrum, it seems as though I favor letter combos that use my index and middle fingers, and that fall on rows I can see (middle and top). Since the bottom row is obstructed, and the edges force me to use my ring and pinky fingers, the bottom edges are the worst keys for me to type.

Remember to take all this data with a grain of salt, since the only user right now is me, and my typing cadence may not reflect the norm. I want to acquire a ton more data by trying this on other people over time, to see if there really is a consistent preference for certain key combos over others. I'd also like to look at longer sequences--maybe letter pairs aren't enough to show evidence of a rhythm, maybe we need to look at triplets or quadruplets. It would also be interesting to look at which keys are most often mistyped and deleted.

Using this research, we can devise a more comfortable layout. Maybe one that also has variable key sizes based on letter frequency, letter position, or even finger dominance (e.g. if we often mis-hit the z key with our left pinky, maybe it should be bigger than the other keys). Or maybe one that has variable textures on keys that are hard to locate while touch typing (many keyboards have that tiny bump on the F and J keys to help people find the home row, but that's it. If we expanded that concept, perhaps we could find keys faster without having to glance at the keyboard). 

Or maybe let's think completely outside the box, and throw out the old keyboard design completely. There is so much wrong with not just the layout, but the entire keyboard form factor. Our needs have grown dramatically and it could use a complete overhaul. One idea I saw recently that I really like: the shift pedal. You literally place a piano pedal under your desk and step on it instead of twisting your pinky out to the side to hit the 'Shift' button. With this pedal, typing uses more of your body and almost becomes more like playing an instrument.

People have experimented with all kinds of out-there keyboard ideas:

A really popular idea is the chorded keyer, which lets you type by pressing buttons simultaneously, like chords on a piano. The motivation behind this is really powerful: a keyboard you can use while walking, without looking, that feels satisfying in the hand.

By building on these ideas, we can create input devices that leverage more of our bodies, or at the very least, more of our hands! Because hands don't just have fingertips; they are also capable of applying pressure, rotation, movements, etc. What if we used a handheld keyboard where certain rotations added punctuation, e.g. a wrist flick adds a period?

Our bodies are incredibly versatile and it's time tech took that into consideration.