Swipe testing data collection - adding keyboard size and placement as a data point #1314
Description
Use Cases
When people use swipe typing, they are usually quite inaccurate, and the challenge lies in discerning clear intent from that unclear signal.
However, I have a strong feeling people are inaccurate in different ways depending on how they're holding their phone. In particular, one-handed mode is probably quite a bit different. I suspect people will get increasingly inaccurate, their thumb reaching less and less for further letters, such as "Q" for a right-handed person, for example.
But I suspect various size/position settings have an effect as well.
Proposal
I think it would improve your data collection if you could allow us to set up the fake keyboard on the data collection site in various ways (most critically one-handed mode).
Then in principle the algorithm could be better tailored to various setups.
As a second, more distant step, once swipe typing clears alpha status, perhaps the system could detect when people make lots of "mistakes"and invite them to participate in a data collection session with their current settings imported, attempting to improve an eventual weakness that person would have unearthed.
References
No response