I'm a Big Fan of Multimodal Input and Converting Modes a Little Bit

I'm a big fan of multimodal input and being able to convert modes a little bit. I agree that right now, both the design conventions have not yet evolved to cover even a tiny fraction of what's going to make truly fluid inputs work well. There are cases where, for example, we needed to detect whether the virtual keyboard was on screen versus a hardware keyboard. And you can never know which of these the user has or does not have. So designing interfaces and workflows,. which can adapt to these, are just not there in terms of design patterns and evolutions.