What's magnificent about Dasher is that it works with so many modalities, like eye tracking, suck tube, pressing a single button and waiting, etc. It's able to extract the maximum number of bits of text per minute from the minimum amount of bits of input per minute, harnessing every input device to its fullest potential.
It also seamlessly supports multiple languages, it can be pre-trained on examples of what you write so it adapts to your vocabulary and gives your favorite words more shelf space on the Library of Babel, and you can include symbols and control functions too. You could even have an "emoji" branch where you spell out the :smile: codes, constrained to just the defined codes so it's easier to select them (and see the growing emojis as you get near to spelling them out all the way), incrementally learning your favorites and making them easier to pick.
It drives me nuts how we blindly accept these ancient, kludgy input methods -- QWERTY is the biggest offender, but let’s not forget those shitty swipe keyboards that think gliding your thumb over letters is somehow “intuitive,” or the garbled cell-phone autocomplete systems that keep mangling your sentences until they’re barely recognizable, then texting embarrassing obscenities to your mom.
Dasher is in a league of its own. Instead of shoving you into a historically fixed grid of scrambled letters, or forcing you to hunt and peck through a list of predicted but unpredictable words, Dasher treats writing as a continuous journey through a probability landscape! You don’t tap tiny squares or hope the AI guesses what you meant, you literally “dive” into the next letter or word that’s statistically most likely. Maximum text output for minimum physical effort, without any of the guesswork or awkward ergonomics baked into hunt-n-pecking QWERTY, slippery swipe tricks, or psychotic autocomplete engines.