> By comparison, able-bodied people close in age to the study participant can type on a smartphone at about 23 words per minute, the authors say. Adults can type on a full keyboard at an average of about 40 words per minute.

I had no idea average typing speed is this slow, though the extreme slowness of touchscreen typing is no surprise.

Ah, so it's previous record for BCI typing speed, not "typing speed" in general. Original headline (on IEEE side) is misleading.

Currently to draw a line in AutoCAD, I have to either move the mouse to a "line" button and click it. OR type a command "line" if my hands are on keyboard. Now If I had a direct interface with the machine, just the the thought of "Draw a line" and software is line-drawing mode. Add eye tracking to it, and I simply look at the point on the screen when line shall start and say in my head "here"...... opportuties are endless.

Computer will have difficult time catching up with the amount of commands we are capable of issuing to it. BRING IT ON.

It was 2006, when I did Speec-To-Text data entry of 200+ mobile numbers using good old Windows XP. Progress seems to be have been too slow since than.

But comparing speeds isn't really necessary or even relevant. For a person that literally has no other way to communicate nobody will be complaining the can type 20% slower than a person with a smartphone. The fact they can type at all is important.

Now, for the speeds, I type 80wpm/400cpm but what is more important I can also freely think while I do so. I don't even know consciously where the keys are on the keyboard, I need to focus on a character and then see where the finger goes as if I was bystander, a passenger of a mech body that observes it do something.

I can only assume that writing with brain-computer interface requires incredible focus and most likely engagement of visual cortex which would preclude from using it for anything else (so no ability to imagine any imagery while typing).

Also, Macross Plus, spoiler, for the subconscious throwing up images in a high stress environment; https://youtube.com/watch?t=2077&v=Fgg6p9gSeR0

A buddy of mine used some timing app for our typing speeds, you had to backspace repeatedly to retype a word if you misspelled it. I argued that is unrealistic, you can just use spell check or just catch the typos during your editing read through.

Use hypnosis.

You can have effective fast communication to the machine without any fancy technology (and certainly without surgery!) The subjective experience is "think to type". I expect that these days you can use a camera and ML to "read thoughts" directly from facial muscles, no need for GVR or IMUs on the fingers.

It would be really cool to have a BCI that runs off mental-verbalizing and have a chopper style rapper use it (eg. Busta Rhymes). Imagine people doing transcription learning chopping as skill to get an edge on their job!

What could be extra cool is just imagining a loop construct in your head and it just appears in your text editor.

Is there already a scene for this? Sign me up.

Other tricks we use for noisy touchscreen input could apply here too. For one thing, we've got different models for to collect input. You could try offering autocompletions/predictions to make typing long words faster (long a thing in e.g. single-switch input tools), if watching a screen doesn't slow the person down much. You could try a swipe-like flow where you collect a chunk of data imprecisely but fast (mentally type or scribble a whole word, say), then offer choices.

(Thinking about other extremes, I wonder if there's a way at faster input using a T9-like reduced alphabet (3x3 grid, pick row and column?) or something like that. Or if it could someday work to for people to try to speak or visualize words instead of handwrite.)

You might be able to glue the letter decoding and language model together more closely--feed the letter-decoding NN's uncertainty and less-probable guesses (25% chance this 'e' was really an 'l') to the lang model.

You could learn a person-specific language model, seeding it with a person's writings/speech before the injury or disease (if available), whatever else you think they might need (family/friend names, care requests, etc.), and training the language model as they use the interface (paper already does that with the character-recognizing model).

You could do explicit "mutual training"--while the machine samples how you write letters, it can show you its certainty scores (based on the model it has so far) or maybe something graphical to help you write the letters how it expects (exaggerate certain differences etc.). They already have an "optimized alphabet" that maximizes the machine-visible differences between letters.

From their paper, the existing language model already did very well at producing clean results in this test, but the more you can refine the cleanup, the more you can potentially sacrifice cleanliness of input for greater speed, and maybe get closer to speaking rate.

FWIW some Googling found http://web.stanford.edu/~shenoy/GroupPublications/WillettEtA...

However, considering how much the now-normalized ill effects of the digital privacy dystopia we’re already living in would be multiplied by that development, I really fear for the future.

Sir Francis Galton pioneered the study of psychometrics which showed that intelligence is positively correlated with reaction speed. So it's not surprising that the average person is slow at a cognitive task like typing..

https://en.wikipedia.org/wiki/Francis_Galton

https://en.wikipedia.org/wiki/Psychometrics

https://en.wikipedia.org/wiki/Jensen_box