One of the biggest innovations in smartphone keyboards since the capacitive touchscreen is gesture typing. For those not familiar, keyboard apps like Swype let users drag their fingers from button to button without lifting their finger, essentially connecting keys in order to form the intended word. swiping keyboard gif

As I watched Sywpe’s demo videos in anticipation for the official iOS release, I become eager to try it myself. Not only did it look smooth and easy, but it made sense: it drastically reduces the movement of your fingers when typing. While traditional keyboards, require 4 separate taps to spell out HELLO, a signle continuous, seamless swipe is sufficient for swiping keyboards. 
It sounds great on paper and it looks convenient on video but the reality is that the overwhelming majority of people don’t use them. Matter of fact, when I ask people as to their keyboard preference, most flat out hate it and few say it’s ok but prefer traditional keyboards. 
While I haven’t conducted scientific research, my theory is:
Swipe-based keyboards add extra cognitive burden to the typing experience as they initiate you in a physical finger dance which you have to think out in advance as you plan out the trajectory of your finger while your spelling out the word in your head. And if you take too long to connect the next letter in your word sequence, swipe keyboards will often error (although Microsoft’s Wordflow is pretty good at handling this.)  
When typing on traditional keyboards, you first have to be cognizant of the actual word you wish to type, then visually identify the first letter on the keyboard and finally press it. Since we type so often, we know the general area of the keyboard where the desired letter may be located, which expedites the process as we narrow down our visual range. It’s a tic-tac process that happens in separate stages. 
On the other hand (pun pun) swiping forces you to think of the upcoming letters as you guide your finger. The 2 cognitive processes run simultaneously which burden our mind. While this may not be a significant cognitive increase for short words like “hey” it significantly increases for words like “ephemeral” or “indoctrination.” 
This phenomenon of cognitively burdening the user is also experienced in an analogous computer interaction field: speech-as-input. I cannot count the number of times when I’m dictating to Siri, I pause for a second to think of my next word or phrase and I get cut off as Siri runs a search on my unfinished sentence. That’s because once again, the user is burdened with pre-planning an exact sentence before speaking. In reality, this is not how we naturally talk. We don’t have every word planned out for the remainder of every thought. We take pauses, we ponder, sometimes even stutter. We start a sentence before we know how we’re going to finish it, and that’s precisely why voice as input has never taken off the way Bill Gates swore it would.