a massive pain in ass with these input systems is because they require custom hardware... [is shit on qwerty hardware imo]... this means u have to create your own whole system for alphabet input. replace normal typing for the hardware you go with. this is pain in the ass caus qwerty layout on qwerty keyboard can do decent speed already [~80wpm on qwerty is comfortable enough...] 80wpm on e.g. hardware with 1 key for each finger = 10 key total is interesting problem. 5 keys on each hand means if you break into left-right presses you have 5*5 options = 25 options. 26 letters in alphabet * 2 for capitals + ~20 symbols = ~80 symbols need to be able to input yet you only have 25 options for an ~ideal 2press lefthand-righthand. so you stick etaoinshldr (most frequent letters in english alphabet) on the best/most comfortable keypress combinations for your hardware, then make the other symbols etc. 3press combinations? its a system that needs careful design. i made some SHITTY input system with this input layers/input chains idea years ago and got it to 50wpm then threw in trash... but that wasnt 1 key for each finger... and it was on qwertykeyboardhardware.

so is pain in the ass any input system you make you probably want to not use qwerty hardware and if you want advanced system you still want alphabet input so have to design something for it. i wouldnt advise trying to design for above 50-100wpm.... text input speed shouldnt be big factor of how slow/fast inputting into computer is with these systems unless you're trying to replace stenography. i wouldnt advise chording in general... just bias but i think chording-systems like stenography are generally kind of trash compared to single-input chains but whatever it depends entirely on the frequency distribution and total number of pieces of the data you're trying to input e.g. letter 'e' in english language comes up 7% of time in all text or something really highest character... out of ~80 characters maybe. so entire system has to be designed for the total number and expected frequency distribution of those pieces of data. and expected frequency distribution could be unknown or change depending on what you're building for

generally just think chording is slower when you could break it into 2 presses instead of pressing multiple things at once. get more options that way and is more comfortable probably. depends entirely on hardware... id be more bias to single press chains probably but yea with some hardware chording could be dank i guess... dont know id prob try do single press chains instead

other weird ideas are press strength... e.g. a midi key on weighted midi controller [piano keys].... a really fucked up idea is have the time inbetween keypresses trigger different chains. this one sounds nightmare to learn but gives higher amount of data per second or whatever... i probably wouldnt try to max out somethin that extreme... id prefer build modular reusable easy-to-learn modules that less likely to completely fail/really hard to learn/dependent on timing/dependent on press strength... in general if designing these input systems id be favoring modularity and user comfort and not spending 200 hr on design of layers... caus it easy to get stuck redesigning layers for 100 hrs with ridiculous complex frequency distribution ideas of what would be most comfortable or most speedy...

quick and dirty easy to learn modular things that can be reused if possible and easy-enough muscle memory is sort of goal id wanna take building these input systems in future. old muscle memory can probably be phased out by superior module replacement in future if required if you build with modularity from the start then just replace modules if u really want to later and that module could have more optimizations or more comfort...

hard to give much advice bout building input systems like this caus so many different designs can come up with. and dependent on whatever hardware you can figure out. and hardware could be point of failure if it just doesnt work as expected at speed...

i think the other reason i didnt like chording was it adds more time to muscle-memory learning... and that it wasnt comfortable. idk i was interested in chording even tho looks like it has pitfalls never rlly messed around with chording... i think in general the single press is faster but there might be certain total-amounts-of-end-of-chain/frequency distribution where chording might be better... even if it takes longer to learn. not sure

fun fact be input system where u can add layers to it on-the-fly using the input system itself... this is very complex input system fun fact... and not that useful caus any layers you add in = new muscle memory to learn. but it would be standard in any input system both complex and advanced. theyre not really static code for input system imo... the code would change every day of using it with more layers always being added and layers that fall out of muscle-memory being deleted. theoretical

in above image... whenever end of input chain is reached.... when text is inputted into computer.... system automatically goes back to "main" layer. can have multiple main layers etc. as complex as needed

i also wonder bout rolling motion. rolling fingers from pinky to thumb and omitting certain finger-presses. dont know never made anything like this. made some crappy voice-to-computer one... was alright

modern voice-to-computer you can say '125626346436' as fast as you want and 99.9% it will get it accurate once you get used to the voice-to-computer software you use... not bad way of harvesting human input into computer... not rly as fast as finger movements probably. maybe combining multiple types of user input but then make system more complex. an ideal hardware might be each finger has 10 options instead of 1 option. 10 tiny little keys or 10 hardnesses or 10 motions or something idk... think thered be a way to get more than just binary input from each finger but would make layers even harder to learn into muscle memory. probably not easy to build input systems

id probably try use asetniop 'preliminary typing gloves' metal plate keyboard as hardware. its some guy's cheap method of building hardware for chording-input-system. might work for this single-press type input system https://www.youtube.com/watch?v=LDqbPhY-A4w

then if you learn muscle memory 100 200 hr. then change something. u dont want have to relearn new muscle memory. unless already similar to other muscle memory you already learned. muscle memory transfers between input systems if theyre similar enough, perhaps in some cases so eh. and then how optimized do you make the layers for better press combinations or less press combinations. theres a lot of design work just in designing layer structure. more complex layer structure can increase muscle-memory-learning time

if u wanna do anything massive and practical with these input layers the input system get fairly massive little bit complex etc and then what are you doing? spamming macros into another piece of software? or rebuilding the software you want to use as well... or creating your own software to use with input layers? yrs of work possibly

theres a user on github called 'carptunnel' has code uploaded for the piano inputter i made... from memory it was pretty rough code just to get it working and im not certifi programmer

never rlly used one for practical purposes. made a piano note inputter a year ago that was maybe useable. built it for qwerty tho and found qwerty doesnt work dat well for these input layers.... might build another one later. cbf providing any proof of whether these input systems work. pretty sure they do caus ive done some inputting with them over yrs and then threw in trash