Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.programming > #14340 > unrolled thread

How is the CharaChorder keyboard even possible?

Back to article view | Back to comp.programming

Contents

  How is the CharaChorder keyboard even possible? Themi <tcm4760@gmail.com> - 2021-12-17 00:44 -0600
    Re: How is the CharaChorder keyboard even possible? Julio Di Egidio <julio@diegidio.name> - 2021-12-17 00:38 -0800
      Re: How is the CharaChorder keyboard even possible? Themi <tcm4760@gmail.com> - 2021-12-17 15:09 -0600
        Re: How is the CharaChorder keyboard even possible? Julio Di Egidio <julio@diegidio.name> - 2021-12-18 02:39 -0800
    Re: How is the CharaChorder keyboard even possible? root <NoEMail@home.org> - 2021-12-18 05:25 +0000

#14340 — How is the CharaChorder keyboard even possible?

Recently I stumbled across a video of someone typing very fast on a 
standard query keyboard using the "charachorder lite" keyboard. Now, 
despite only typing at about 60wpm I am obsessed with finding ways of 
typing faster so this really caught my eye. https://www.charachorder.com/

This keyboard uses "chording", meaning you can press more then one key 
at a time in order to type out a whole word. The steno machines used for 
court reporting and live captioning also use chording, I've never tried 
to learn how to type steno though since it seems extremely difficult and 
steno machines are crazy expensive.

Anyways, the charachorder claims to having a more simple chording system 
that is easier to pick up. You're supposed to be able to click on every 
key making up a word at once and it will type the whole word out, that 
way you don't need to type every letter in order individually. Naturally 
my first thought was "I want to make a program that does that for 
myself", and then my second thought was "wait what about anagrams??".

Imagine you want to type the word "this", so you press keys "this" at 
the same time. How can the keyboard possibly know if you intended to 
type "this" or if you intended to type "shit"? The only answer I can 
think of is context, maybe there's some sort of predictive ai that's 
able to tell in context if you likely meant this or shit. At the same 
time though, how many words can you make with the letters "t h i s"?

Well I downloaded a word dictionary and wrote a short script that 
figured it out. I came up with the following list of words: ['hist', 
'hists', 'hits', 'isth', 'shist', 'shists', 'shit', 'shits', 'sisith', 
'sith', 'stith', 'this', 'thitsi', 'tshi', 'tsitsith']

I tried a whole bunch of other common words and nearly every time there 
where 5+ fairly common words that fit the same chording.

I can't imagine a predictive ai could be very accurate at guessing 
between what of these words you intended to type! Maybe I'm 
underestimating ai, or maybe I am missing something here. There's got to 
be some sort of big caveat here.

Unfortunately I can't take a peek into the charachorder's software since 
the keyboard has a microprocessor that does all the work built into it. 
I may try to learn how to make a predictive ai in order to test this out 
since I've just got to know soon.

What do you guys think? Am I missing something obvious here?

[toc] | [next] | [standalone]

#14341

On Friday, 17 December 2021 at 07:44:26 UTC+1, Themi wrote:

> Recently I stumbled across a video of someone typing very fast on a 
> standard query keyboard using the "charachorder lite" keyboard. Now, 

Sure, nice advertisement...

> myself", and then my second thought was "wait what about anagrams??". 

At a quick glance, looks like the T9 for mobile phones or e.g. search suggestions when you start typing in the browser, and similar: it also means it has to keep some kind record of past entries, i.e. of anything typed including logins, passwords and bank account numbers, so 'd be more curious to know how they handle that.

That said, there certainly is room for creative approaches to computer keyboards: it is well known that qwerty and similar are the opposite of ergonomic/effective...

Julio

[toc] | [prev] | [next] | [standalone]

#14343

On 12/17/2021 2:38 AM, Julio Di Egidio wrote:
> On Friday, 17 December 2021 at 07:44:26 UTC+1, Themi wrote:
> 
>> Recently I stumbled across a video of someone typing very fast on a
>> standard query keyboard using the "charachorder lite" keyboard. Now,
> 
> Sure, nice advertisement...
> 
>> myself", and then my second thought was "wait what about anagrams??".
> 
> At a quick glance, looks like the T9 for mobile phones or e.g. search suggestions when you start typing in the browser, and similar: it also means it has to keep some kind record of past entries, i.e. of anything typed including logins, passwords and bank account numbers, so 'd be more curious to know how they handle that.
> 
> That said, there certainly is room for creative approaches to computer keyboards: it is well known that qwerty and similar are the opposite of ergonomic/effective...
> 
> Julio

I'm definitely not advertising haha, I'm questioning if this product 
even works. I found it on kickstarter which makes it super questionable 
in my opinion since I've seen so many projects from there fall on their 
face in the past.

hmm that is an interesting point, it probably does keep a history of 
past entries. This probably isn't something that warrants using ai then 
now that I think about it, I could certainly try and use a simple 
"frequently used" algorithm that maybe takes into account the context of 
the previous few words like suggestions on mobile phone keyboards do. I 
think I may try that out.

Maybe using standard chording software where you define what keys 
correlate to what words more of the move here since you won't end up 
typing words you didn't expect to like it seems like you may using a 
system like CharaChorder is.

[toc] | [prev] | [next] | [standalone]

#14345

On Friday, 17 December 2021 at 22:10:00 UTC+1, Themi wrote:
> On 12/17/2021 2:38 AM, Julio Di Egidio wrote: 
> > On Friday, 17 December 2021 at 07:44:26 UTC+1, Themi wrote: 
> > 
> >> Recently I stumbled across a video of someone typing very fast on a 
> >> standard query keyboard using the "charachorder lite" keyboard. Now, 
> > 
> > Sure, nice advertisement... 
> > 
> >> myself", and then my second thought was "wait what about anagrams??". 
> > 
> > At a quick glance, looks like the T9 for mobile phones or e.g. search suggestions when you start typing in the browser, and similar: it also means it has to keep some kind record of past entries, i.e. of anything typed including logins, passwords and bank account numbers, so 'd be more curious to know how they handle that. 
> > 
> > That said, there certainly is room for creative approaches to computer keyboards: it is well known that qwerty and similar are the opposite of ergonomic/effective... 
> 
> I'm definitely not advertising haha, I'm questioning if this product 
> even works. I found it on kickstarter which makes it super questionable 
> in my opinion since I've seen so many projects from there fall on their 
> face in the past.

I find the very mission objectionable: "to elevate average human text transmission speed (40 wpm) above and beyond average human text comprehension speed (250 wpm)".  Why do we even need that if not along the way of the extinction of thinking?  More concretely, isn't at that point rather speech recognition and beyond the way to go?

> This probably isn't something that warrants using ai then 
> now that I think about it, I could certainly try and use a simple 
> "frequently used" algorithm that maybe takes into account the context of 
> the previous few words like suggestions on mobile phone keyboards do. I 
> think I may try that out. 

That's often good enough already.  Yet don't discount ML as it's doing great in the realm of next word prediction up to automatic writing, with very good results over relatively limited computing resources.

> Maybe using standard chording software where you define what keys 
> correlate to what words more of the move here since you won't end up 
> typing words you didn't expect to like it seems like you may using a 
> system like CharaChorder is.

By the videos, it rather seems it's exactly like changing a word in T9.  Rather, good luck learning yet another "bag of tricks".  And when you borrow your friend's PC, back to square one?  But I guess a subcutaneous chip with Bluetooth interface connecting to an onboard universal device might solve even that...

Julio

[toc] | [prev] | [next] | [standalone]

#14344

Themi <tcm4760@gmail.com> wrote:
>
> What do you guys think? Am I missing something obvious here?

You clearly demonstrated one proof that typing all the keys
of a word simultaneously doesn't work. However, typing two
keys at once would allow them to represent 676 words. Too
much to have to remember. There are 44 phonemes in English.
These could be remembered and used to represent words which
an intelligent program to transform back to standard spelling.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.programming

csiph-web