In case someone unfamiliar finds this post, in March, HeliBoard began its project to create a FOSS gesture library. This is the feature which allows users to swipe their fingers across an on screen keyboard, in order to produce words. Currently—if you are using this feature at all—you are utilizing a proprietary library, such as from Google.
Right now, it's in the collection phase, as this will need a mountain of user data to analyze before anything substantial can happen. This phase is planned to last until several months from now, and the more people send in words, the better!
Info on the project: Video (PeerTube) | Text Tutorial
Maybe seeing someone who's serious about sending in data will inspire others to send some too. Here's hoping, anyway! I've been doing a fair bit, but a handful of people doing a lot less would easily exceed my output. Ape together...strong!
I'm not keen on typing on mobile at all, but if I have to, I'd much prefer to do gesturing. It can be rather annoying to see if mess up a lot of course, and part of this project is hopefully to have a base which can improve upon what already exists. I don't know how true that will be, but I'm hopeful, and the project lead who got the grant seems optimistic!