Author Topic: Balanced Keyboard Layout  (Read 41836 times)

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #950 on: October 03, 2017, 02:23:42 PM »
For a Planck, I'd be curious for matrix layout that assigns just 2 letters, one for each thumb.

I tried quite a few layouts like that (E/T/A/N etc) without success, except for single T or H. So was surprised that Den's 3-letters work so well.

My famous overkill mockup is here: http://iandoug.com/?p=183  (click image to enlarge it)

sdothum

  • Member
  • **
  • Posts: 22
    • View Profile
Re: Balanced Keyboard Layout
« Reply #951 on: October 03, 2017, 03:06:15 PM »
I tried quite a few layouts like that (E/T/A/N etc) without success, except for single T or H.

Interesting.

Well, I guess I that means I'll be settling on BEAKL 8. I must be one of the minority here prefering 8 (for its English prose ranking) over 9 -- KH on the home row is genius.. Q, on the other hand, seems weirdly out of place to me, lol).

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #952 on: October 03, 2017, 03:57:53 PM »
@Den

Um isn't "same hand" getting punished going to be opposite of "rolls get rewarded"?

Am playing around with Essie style, and vs for example you initial version (BEAKL CPL 0), have lower same finger, lower distance, lower finger usage, but higher same hand, and overall score not good enough ... currently on Den1 I'm midway between yours and Seelpy 1.22, and on DenKLatest, worse than CPL 0 and Seelpy 1.5.

So trying to understand why. At present pondering if much higher same-hand usage is getting punished too much.

Layout is still very much a work in progress. Initial idea was to hardly use pinkies at all by putting extra row at top, but leads to serious issues resolving same-finger issues. At present about 4% usage each on pinkies, on Putin. Code would probably be worse.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #953 on: October 03, 2017, 06:42:33 PM »
High same hand is unavoidable due to excessive use of thumb modifiers. But acceptable for this purpose:
1. Other stats are low to compensate.
2. Half the same hand combo is with thumb, the strongest finger.

Scores categories are independent. Rolls will analyze the quality of rolls, including perhaps row jumping.
Support me on Patreon

I saw. I conquered. I came.

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #954 on: October 04, 2017, 07:11:40 AM »

sdothum

  • Member
  • **
  • Posts: 22
    • View Profile
Re: Balanced Keyboard Layout
« Reply #955 on: October 04, 2017, 10:17:11 AM »
See comments.
http://xahlee.info/kbd/char_frequency_counter.html

Interesting.

The H reference certainly pertains to my preference for BEAKL 8 over 9 for English prose. Certainly for me, layout optimization for programming is unimportant -- layers can more effectively address specific language and workflow needs. Plus, coding for me is probably 97% thinking and 3% actual typing, and I have written a lot of lines of code in my time.

I've read similar views regarding the cost-benefit of learning supposedly "better" layouts, especially when individuals factor in their actual time typing on the keyboard. Except for writers, data entry clerks and people with physical issues, cost-benefit is probably a questionable criteria because there is arguably no impactful return. Heck, there's probably a segment of the population who only interface with their touch phones.

For me, it ultimately comes down to the pleasure principle: is the layout enjoyable to type on? Finger memory plays a huge role. I will always find Colemak a pleasure to type on with its long finger rolls despite it not necessarily measuring the best. That's why vast numbers of people comfortably use QWERTY. I'm always surprised at people who can type at lightning fast QWERTY speeds I can't even imagine ever typing at, complaining about the difficulty of achieving and surpassing it with a "better" layout.

Probably, the faster you can type, the more finger memory there is to overcome. Once that finger memory is ingrained, metrics are rendered rather moot.

Anyway, Ian, I am probably way off the mark on your thread..

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #956 on: October 04, 2017, 12:39:19 PM »
Cost benefit in terms of what? Monetary? Health? Pleasure? Progress? Whereas almost every other field of ergonomics has embraced better efficiency, thus progress and innovation, keyboard layouts are destined to stagnate. Cynics might blame this on lack of public interest, or ignorance.

Knowledge and curiosity are itself benefits. Unfortunately there are systems we know are inefficient, yet we must to continue to bear with it indefinitely. Long standing traditions are immovable objects. Yet doesn't mean we should stop trying to change the world for better, or stop seeking the true. (The One True Layout (tm))*

[searching one true layouts gives me bunch of stuff about CSS layouts.]
« Last Edit: October 04, 2017, 12:42:24 PM by Den »

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #957 on: October 04, 2017, 01:21:54 PM »
Myself, Ian be INTJ and our motto is "everything can be improved".

Including keyboard layouts. I hate bad design and QWERTY is so bad that it deserves to die. Humans deserve better and we're smart enough to come up with something better. ("we" as in humans).

FWIW I came across Lillian Malt's article today. (Malt being the designer of Maltron layout, and co-incidentally born in South Africa... Despite it's age it still holds up well in assorted tests.)

Attached. She also pleads for something better than QWERTY.

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
X7
« Reply #958 on: October 04, 2017, 02:14:55 PM »
I wasn't getting anywhere fast with Seelpy/Essie new version, so I switched back to X6.4H layout and saw what I could do on Den KLatest....

Attached layout does quite well... wins some (yes, even beating P_NR), second in others (usually to P_NR). So far have only attended to letters and basic punctuation, so numbers and other puncs are all over the place. In spite of that it still does well on Keyboard Layout Editor test.

Does not do so well on Den1 scoring though.

Enough for one day :-)
Screenshot and json attached.

sdothum

  • Member
  • **
  • Posts: 22
    • View Profile
Re: Balanced Keyboard Layout
« Reply #959 on: October 05, 2017, 05:52:21 AM »
Den:

I downloaded your beakl30.txt (opt config) you provided last week but opt is not generating anything closely resembling the BEAKL 8 layout -- when using the stock opt wordlist and bigram/trigram files provided with that app.  I'm assuming it's these sources that account for differences.

Could you provide these lists and *gram files that you run against your config? Just trying to replicate your results to gain a deeper appreciation of your work and opt.

mstacker

  • Member
  • **
  • Posts: 27
    • View Profile
Re: Balanced Keyboard Layout
« Reply #960 on: October 05, 2017, 04:16:47 PM »
If I could do that it would save me a fortune on having keycaps printed... (have them done by laser).
Not sure I'm that talented... :-)
You know. When I first started learning dvorak simplified, I procured some small letter stickers that I used to label my keycaps.  Something of this ilk would likely serve the purpose well. I believe the ones I chose where dvorak keycap labels. There is no reason you cannot place them in any order you like.

https://www.amazon.com/SIMPLIFIED-KEYBOARD-LETTERING-TRANSPARENT-BACKGROUND/dp/B0011FI3O8/ref=sr_1_1_sspa?ie=UTF8&qid=1507245361&sr=8-1-spons&keywords=dvorak+keyboard+stickers&psc=1

Incidentally, I've been typing with my modified beakl layout keyboard for about five days now. This message took way too long to type.

Still, kinda cool,

Matt

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #961 on: October 05, 2017, 05:03:27 PM »
You know. When I first started learning dvorak simplified, I procured some small letter stickers that I used to label my keycaps.  Something of this ilk would likely serve the purpose well. I believe the ones I chose where dvorak keycap labels. There is no reason you cannot place them in any order you like.

https://www.amazon.com/SIMPLIFIED-KEYBOARD-LETTERING-TRANSPARENT-BACKGROUND/dp/B0011FI3O8/ref=sr_1_1_sspa?ie=UTF8&qid=1507245361&sr=8-1-spons&keywords=dvorak+keyboard+stickers&psc=1

:-)
Gotta love Amazon: "This item does not ship to South Africa."

I have a Brother label printer ( PT-1230PC) that I used to print stickers. They tend to start moving after a while. Or "split" so that the printed layer comes off, leaving a layer with glue on the key. Hence my question originally, looking for a better version.

I think originally I resorted to handwritten letters on masking tape ....
Remainders from when I was testing Workman Programmer or somesuch attached...
Clearly JQZX are the least used letters... :-)



Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #962 on: October 05, 2017, 07:45:37 PM »
Den:

I downloaded your beakl30.txt (opt config) you provided last week but opt is not generating anything closely resembling the BEAKL 8 layout -- when using the stock opt wordlist and bigram/trigram files provided with that app.  I'm assuming it's these sources that account for differences.

Could you provide these lists and *gram files that you run against your config? Just trying to replicate your results to gain a deeper appreciation of your work and opt.

my corpus is huge file with mix of stories, forum posts, code, custom magic cards, etc. see attached.

See comments.

http://xahlee.info/kbd/char_frequency_counter.html

Quote
ETAOIN Does Not Represent Common Text

If you are designing a keyboard layout, then ETAOIN is not a good model to use.

ETAOIN does not contain H.

H is used in lots common words, and these words are used frequently...

Vast majority of keyboard users, are likely to type just common chat text only, not printed books text, such as novels and academic text.

He overstates the argument for H, which is over-represented in prose and proper grammar. In fact, common chat text and social media deemphasizes H. Ppl tend to simplify grammar by omitting articles and helping words, like: the, that, with, what. either dropping those words or replaced with acronyms, like: WTF, BTW, w/. 'TH' often replaced w/ 'd'/'da': da, dat, dem. 'WH' could drop the H: y, wat, wen

The frequency of letters in my corpus is:
Code: [Select]
etaronislhcdmu.pgy/f
After comparing many, many, many corpus of many different topics, sources, platforms, vernacular, etc., ETAONI is still consistently the top frequent letters. In fact, recent texts tend to show more R and S, and less H.
« Last Edit: October 05, 2017, 07:54:49 PM by Den »

sdothum

  • Member
  • **
  • Posts: 22
    • View Profile
Re: Balanced Keyboard Layout
« Reply #963 on: October 06, 2017, 05:56:04 AM »
my corpus is huge file with mix of stories, forum posts, code, custom magic cards, etc. see attached.

He overstates the argument for H ... recent texts tend to show more R and S, and less H.

Thank you Den for the bigram/trigram lists. I am still unable to come close to reproducing your results, so I will need to follow your lead :-)

How much hand tweaking of the layouts did you have to do to arrive at BEAKL 8, if at all? Regardless, however your magic, I am quite enjoying the layout -- when the wall of inapplicable finger memory doesn't rear its head.

As for "H", letter frequency materials I have referenced place "H" right behind "SR" in usage and the "TH/THE" digraph/trigraph reign number one. That is why BEAKL 8 appeals to me, its scoring on prose bearing this out.

As someone who still loves to read and write, I am horrified at the thought keyboard layout design will be impacted by the devolution of English language usage to the tribal grunts of social media. Let these people continue to use QWERTY, I say! (Though, diminishing discourse aside, touch phone technology obviously plays an increasing role in this linguistic trajectory.)

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #964 on: October 06, 2017, 06:09:25 AM »
As someone who still loves to read and write, I am horrified at the thought keyboard layout design will be impacted by the devolution of English language usage to the tribal grunts of social media. Let these people continue to use QWERTY, I say! (Though, diminishing discourse aside, touch phone technology obviously plays an increasing role in this linguistic trajectory.)

We've got a bit of a side thread on the desirability/viability of switching to phonetic rather than conventional spelling, using IPA or something else.

From my point of view, that will require some changes to English to get rid of words that sound the same (eg ate eight), and even better, same words used for different things (eg stick, lead, etc)

http://shenafu.com/smf/index.php?topic=189.0

work in progress:
http://www.keyboard-design.com/pangalactic.html



sdothum

  • Member
  • **
  • Posts: 22
    • View Profile
Re: Balanced Keyboard Layout
« Reply #965 on: October 06, 2017, 07:28:42 AM »
We've got a bit of a side thread on the desirability/viability of switching to phonetic rather than conventional spelling, using IPA or something else.

Interesting.. phonetic/digram/trigram shortcuts could certainly be implimented easily enough on a platform that has layers, like a Planck.

Though, quite honestly, I actually enjoy the act of typing -- the feel of it. Towards a more phonetic approach to typing, I would be more inclined to invest time in  learning Plover (steno) and its key chord approach to typing which seems more efficient (to me) -- though, is a real mind twister with a huge learning curve (not to mention suitable keyboard to type on)!

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #966 on: October 06, 2017, 07:32:27 AM »
Though, quite honestly, I actually enjoy the act of typing -- the feel of it. Towards a more phonetic approach to typing, I would be more inclined to invest time in  learning Plover (steno) and its key chord approach to typing which seems more efficient (to me) -- though, is a real mind twister with a huge learning curve (not to mention suitable keyboard to type on)!

I looked at Plover and asked how I would write a program with it... :-)

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #967 on: October 07, 2017, 05:42:52 AM »
Regarding the idea of a single, easy to peruse source of information for all the BEAKL layouts. Could probably create a separate page on Deskthority wiki. (instead of starting our own wiki.) And or static page on this site, (with some interactive fun stuffs? )

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #968 on: October 08, 2017, 02:52:20 AM »
Regarding the idea of a single, easy to peruse source of information for all the BEAKL layouts. Could probably create a separate page on Deskthority wiki. (instead of starting our own wiki.) And or static page on this site, (with some interactive fun stuffs? )

DT would probably get you the widest exposure.

Might be worthwhile long term to have your own page too... :-)

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #969 on: October 08, 2017, 04:54:46 AM »
Code: [Select]
BEAKL 10

qhoux gdnmv
yiea. csrtw
j/',z bplfk


This has superb rolls and low same finger.

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #970 on: October 08, 2017, 05:45:21 AM »
This has superb rolls and low same finger.

KLatest please :-)

Thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Ergolinear / Matrix prior art
« Reply #971 on: October 08, 2017, 06:11:00 AM »
In my ongoing quest to explore brave new layouts and boldly go where no keyboard has gone before, stumbled across two things which I suppose would count as "prior art" re Ergolinear/Matrix/Programmer's Keyboard layouts.

Although I did not copy them and had not seen them before I came up with the ideas. I do admit to being aware of Maltron style and probably borrowed some of the ideas and rejected others.

Attached PDF from Turkey discusses 10 patents around keyboards, including Patent 6241406 by Yan, June 2001). (Page 12).
Design is very similar to what I came up with.
Wonder if there are patent issues with Ortholinear re this.

Then we have the Krush 200 project, which seems to have faded. Goal was to have a keyboard capable of doing 200 WPM.
They had fund-raising campaign etc.

Google search, see pages and images results.
http://tinyurl.com/y97hch98

I see this also had the idea of putting common words onto a key ...  I just had "and" and "the" and a few other common things like "ion", Krush had a lot more.
Some of the comments from people dissing his design now seem amusing. Guess we have different frames of reference. I've seen other commentators dissing the big space bar on QWERTY.
Will see if I can put this layout onto Matrix/Ergolinear, without the words.
But t is not on home finger, while l is ... that's not going to do well.


iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Matrix
« Reply #973 on: October 09, 2017, 12:21:51 AM »
@ Den

Um, just to clarify, your Matrix layout is "complete" in the sense that Ergodox layout is complete? Or will a physical version have extra keys (nav/function/numpad/whatever)?

thanks,  Ian

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Matrix
« Reply #974 on: October 09, 2017, 09:31:10 AM »
@ Den

Um, just to clarify, your Matrix layout is "complete" in the sense that Ergodox layout is complete? Or will a physical version have extra keys (nav/function/numpad/whatever)?

thanks,  Ian

You mean add more buttons? ideally there should just be contiguous buttons across all the rows, no gaps. but i need to fix the code to detect missing buttons and fill them in as blank letters rather than skipping them from the configuration page. would make conversion between boards somewhat smoother. (though keys may be moved around, but at least not missing.)

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Matrix
« Reply #975 on: October 09, 2017, 10:12:24 AM »
You mean add more buttons? ideally there should just be contiguous buttons across all the rows, no gaps. but i need to fix the code to detect missing buttons and fill them in as blank letters rather than skipping them from the configuration page. would make conversion between boards somewhat smoother. (though keys may be moved around, but at least not missing.)

Not sure you understood me (or I am not fully understanding you...)

For purposes of my Best Layouts tests, I need to have images of all the layouts, so wrote a script to map KLA layouts onto KLE templates, samples attached.
Done ANSI, ISO and Ergo, now need to do Matrix/Ergolinear. So need to know how the KLE template should look ... somewhat similar to Ergodox or something more slab-like?

Need to redo some of the code to use trig to work out co-ordinates ... at present lots are hard-coded but that gets messy with angled keys and different size keys. Was just worried my sums in PHP and KLE's sums in JavaScript would produce slightly different results.

Had to combine several Noto fonts into one, and add some missing glyphs, to handle boards like ADNW. Doing the modifiers with different font because Noto doesn't have all the Misc Tech symbols needed.
I see some of the Author stuff will need fixing.

Suppose it will provide a useful Internet Keyboard Layout Database when done :-)


iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #976 on: October 09, 2017, 10:21:00 AM »
Incidentally, I've been typing with my modified beakl layout keyboard for about five days now. This message took way too long to type.

Some modelling on how long it takes to learn a new layout.

https://users.aalto.fi/~jokinej10/visual-search/

mstacker

  • Member
  • **
  • Posts: 27
    • View Profile
Re: Balanced Keyboard Layout
« Reply #977 on: October 09, 2017, 07:25:15 PM »
Some modelling on how long it takes to learn a new layout.

https://users.aalto.fi/~jokinej10/visual-search/

Good find. Very interesting.  Not sure what level of proficiency this group considers a keyboard layout learned. Speaking from personal experience, it takes longer than 50 hours to be proficient enough to use a layout at work. Maybe 50 hours, one hour a day or something. That may hold true. I'll let you know in a few months. My past experience switching cold turkey, 8-10 hours typing per business day, it's more like three months. Mileage may vary for others. Would love to hear others experiences.

In any case, great find. Thanks for the post,

Matt

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Matrix
« Reply #978 on: October 09, 2017, 11:45:31 PM »
Matrix should be X rows by Y columns of keys with no gaps. see klatest: http://shenafu.com/code/keyboard/klatest/#/config

Currently 19*5=95 keys. The new additional keys are indexed after the original keys, so it won't mess up the order for existing layouts.

I don't bother what goes on the extra keys outside of the ascii characters that we test. You can set it to whatever you like for KLE.


iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Matrix
« Reply #979 on: October 10, 2017, 12:09:12 AM »
Matrix should be X rows by Y columns of keys with no gaps.

That's the part that confuses me... when you say "no gaps" I presume you mean no gaps on a given hand, but hands are indeed separated.

Thanks for adding BEAKL 10.

Cheers, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Matrix
« Reply #980 on: October 11, 2017, 03:19:01 PM »
That's the part that confuses me... when you say "no gaps" I presume you mean no gaps on a given hand, but hands are indeed separated.

Or not. :-)

Thanks, Ian

mstacker

  • Member
  • **
  • Posts: 27
    • View Profile
Re: Balanced Keyboard Layout
« Reply #981 on: October 12, 2017, 05:51:00 AM »
@Den.

The beakl clp 1 layout looks interesting. I like that it requires a minimum number of keys. Have not yet determined if that is a good thing yet. Also, clp 1 seems to have a really low pinky finger usage for almost everything I feed the kldlatest utility. Any idea where the beakl clp train of thought is heading?

Matt

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Matrix
« Reply #982 on: October 12, 2017, 01:13:10 PM »
Currently 19*5=95 keys. The new additional keys are indexed after the original keys, so it won't mess up the order for existing layouts.

Okay got it working, spent way too much time fighting with the trig, eventually got it to behave (for especially dealing with angled keys).

Just have Ergolinear to do now. Will put my layouts back on Ergolinear with extra central keys.

mstacker

  • Member
  • **
  • Posts: 27
    • View Profile
Re: X7
« Reply #983 on: October 14, 2017, 05:51:33 AM »
I wasn't getting anywhere fast with Seelpy/Essie new version, so I switched back to X6.4H layout and saw what I could do on Den KLatest....
Very cool. The X7.1 layout, that is. More and more I'm liking the idea of placing letters on the thumb keys.

Matt
« Last Edit: October 14, 2017, 09:11:10 AM by mstacker »

mstacker

  • Member
  • **
  • Posts: 27
    • View Profile
Re: Balanced Keyboard Layout
« Reply #984 on: October 14, 2017, 06:17:26 AM »
Myself, Ian be INTJ and our motto is "everything can be improved".

Including keyboard layouts. I hate bad design and QWERTY is so bad that it deserves to die. Humans deserve better and we're smart enough to come up with something better. ("we" as in humans).

FWIW I came across Lillian Malt's article today. (Malt being the designer of Maltron layout, and co-incidentally born in South Africa... Despite it's age it still holds up well in assorted tests.)

Attached. She also pleads for something better than QWERTY.
Interesting paper. Explains well a lot of thought concerning keyboard layout improvement. The metrics Malt uses look similar to what you and Den have been using. Cool.

On that note, rethinking the importance, or lack thereof, of the home row and reducing the use of the pinky fingers is a major innovation in my view.

Good find. Keep them coming.

Matt

« Last Edit: October 14, 2017, 06:25:25 AM by mstacker »

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #985 on: October 14, 2017, 10:51:36 AM »
Good find. Keep them coming.

:-)
Here you go: http://www.datapointed.net/2010/01/typing-styles-compared/

I'm probably close to Von Style, I think.

@Den, doesn't KLA assume each finger returns home UNLESS it is needed for the next key, eg typing "hy"on qwerty?

Wonder that the effect of Robot Style would be on scoring. Not sure it would be comfortable to type like that, your fingers could end up in strange stretches.
Mine hover and move where needed, and I think certain sequences have been optimised over the years, like ".za " which I have to type a lot, the za is middle->ring. (qwerty).


Cheers, Ian


« Last Edit: October 14, 2017, 11:57:32 AM by iandoug »

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #986 on: October 14, 2017, 11:07:10 AM »
Interesting paper.

Attached paper attempted to prove Dvorak better than QWERTY. Their methodology was much simpler than KLA/opt/etc, which is curious. I tried to get an account at some website that publishes academic papers, but since I haven't published anything in one of the cliquey Academic Publications, they looked down their nose at me. But I think what we are doing here should count as research, we're going into new territories. If we were in a Big Corporate I think a lot of stuff would have been patented, if only for bragging rights.

Anyway, their methodology seems to consist of counting row changes, and using that to prove Dvorak better than QWERTY. They look at Hand Movement, instead of finger movement.
Way simpler than KLA.

Source code in Python included.
« Last Edit: October 14, 2017, 11:57:47 AM by iandoug »

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #987 on: October 15, 2017, 02:46:40 AM »
Even the early (before computer-generated) Dvorak is easily 10% improvement on Q. Then the recent layouts discovered here is another 10%-15% gain. When we also improve the physical design (especially the thumb keys), we get another 10%-20% gain. A total 30%-40% more efficiency on archaic ANSI/ISO Q. is not worth the trade-off, they say.

They claim that most people don't even try touch typing. Is that because it's hard for them; or is it because the unintuitive, messy de facto layout (Q) that scares them away? Yet they want to retain a layout that's not even good for hunt and peck method, that untrained people use, for no good reason. Are people really afraid of cleaner, easier, intuitive layouts that benefit both touch typists and hunt-and-peckers?

On that note, rethinking the importance, or lack thereof, of the home row and reducing the use of the pinky fingers is a major innovation in my view.

The discovery of the home block coincidentally also works for touchscreen (and possibly other interfaces). Many of the layouts I designed for mobile devices (see http://shenafu.com/smf/index.php?topic=88.msg308) tend to optimize around circles and squares--which are more ergonomic and efficient shapes than flat rectangles and long lateral distances.


iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #988 on: October 15, 2017, 04:32:49 PM »
They claim that most people don't even try touch typing. Is that because it's hard for them; or is it because the unintuitive, messy de facto layout (Q) that scares them away?

May be related to the fact, that as someone said before here, ANSI/ISO is designed for two right hands.

I know there are Colemak variants that attempt to address this problem on ANSI (eg http://waddles.org/culemak-slight-variation-colemak/) , which is the wrong way to solve the problem.

Maybe if they made the Left Shift shorter, and put Q on the bottom next to Z, and angled the left "finger XX goes here" mapping inwards, they'd have more luck.
This is probably more doable on ISO but the stagger on the top row is still favouring two right hands. Bottom row is *almost* symmetrical. ANSI/ISO stagger is just plain silly on an electronic keyboard. We don't need to worry about typists migrating from manual typewriters any more.

My version of Culemak, fx, would put FRX on ring, rather than WRX.

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #989 on: October 15, 2017, 04:44:57 PM »
They claim that most people don't even try touch typing.

On a related note, my kid may be going to a school where touch typing is on the curriculum. Which surprised me (since there are lots of other more important things for middle school kids to be learning) but naturally I immediately started wondering what would happen if she rocked up at school with a home-made keyboard with a superior layout, and made them all look silly.... :-)
She already types quite fast for "self taught"... and is much faster on the phone with two thumbs than I am with one blunt index finger. I can't get "two thumbs" method to work.

I "learnt to type" playing Zork till the early hours of the morning (yes, on the Atari 800...)
Would have been easier to get into hell if I had been Catholic... book/bell/candle was totally illogical to me :-)

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #990 on: October 16, 2017, 03:17:03 AM »
LMAO. The hunt-and-peck (HP) layouts using BEAKL 8 do better than Q, Dvorak, Colemak (non HP). Even more surprising, Dvorak HP outperforms BEAKL 8 in HP format. OTOH Colemak and Q show minimal change as HP for Classics Collection; they do better on other corpus, but never as good as Dvorak HP.

What is HP format? The strong fingers rest on ASD/KL; (on Q), with pinky only for shift. With both hands diverted to the edges, the index must work more and move farther to reach the inner columns; thus simulating hunt-and-peck.

see attached for HP layout and results (for Classics Collection).

So your touch typing teachers are wrong. The home row is actually Shift-ASD / KL;-Shift (on Q), Shift-AOE / TNS-Shift (on Dvorak),
« Last Edit: October 16, 2017, 03:25:00 AM by Den »

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #991 on: October 16, 2017, 04:55:47 AM »
What is HP format? The strong fingers rest on ASD/KL; (on Q), with pinky only for shift. With both hands diverted to the edges, the index must work more and move farther to reach the inner columns; thus simulating hunt-and-peck.

Co-incidentally I was playing around with Culemak last night, some tweaks got a better score (using Den1 and Putin's speech).
So I loaded your H+P variant, but results are disappointing.
On Den1/Putin, it came stone last. (I'm using Putin because of similarity of letter frequency to Generally Accepted Frequency for English).(screenshot below)
Likewise on Den1/Classics (more annoyingly, MTGap wins this round)
Likewise on Den1/Alice (where my plain Culemak is again on top)

On KLatest, it's best of the 4 Colemaks on Classics (which was encouraging), and worst of the four on Putin.
This sort of variation worries me ... as in, is that how it is supposed to be, or are there subtle issues in the scoring affecting things?


« Last Edit: October 16, 2017, 05:02:03 AM by iandoug »

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
c^2
« Reply #992 on: October 16, 2017, 11:40:36 AM »

Anyone able to persuade Google to show the actual layout?

https://medium.com/@ebclay/the-story-behind-the-invention-of-the-c-2-keyboard-60e6d30eaf5f

I could not find it. Have sent him a tweet.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #993 on: October 16, 2017, 03:26:03 PM »
Colemak HP seems to fail on same finger. Frequent bigrams hit with same index: en, st, he, el, le, etc.
« Last Edit: October 16, 2017, 03:32:54 PM by Den »

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: c^2
« Reply #994 on: October 16, 2017, 06:24:27 PM »
Anyone able to persuade Google to show the actual layout?

https://medium.com/@ebclay/the-story-behind-the-invention-of-the-c-2-keyboard-60e6d30eaf5f

I could not find it. Have sent him a tweet.

I don't think there is a picture because he is trying to patent and sell it as proprietary. Might also come with a new physical design. Given how much we know about the market, his chances of success are very slim. He'd be better off disclosing his ideas on geekhack, observe the amount of interest, and try something group sale like massdrop, kickstarter,

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: c^2
« Reply #995 on: October 16, 2017, 10:22:09 PM »
I don't think there is a picture because he is trying to patent and sell it as proprietary.

Some places I've found quotes describing him as already having patented it (and started/registered a company to make it).
However Google won't find any such thing linking various versions of his name, keyboard/layout/c^2/csquared/c-squared/etc, not even on their patent site.
So maybe he used different names in the patent.
Patent site has over 2000 patents with keyboard/layout since around 2013. Most not relevant to our interests. Will take a while to wade through them all. Maybe should start at 2016 and then work backwards if necessary.

I find it odd that someone with that level of IQ (over 200, apparently) would not have at least floated ideas on GH/DT//r/ etc. Or maybe he used another name, or just lurked. But given that he's asking questions in places like Quora, he did do some online research "publicly" ...

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1039
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #996 on: October 17, 2017, 09:05:51 AM »
KLAtest changes:

-- heat maps of all layouts all at once
-- removed table of key frequency

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #997 on: October 17, 2017, 10:19:53 AM »
KLAtest changes:

-- heat maps of all layouts all at once
-- removed table of key frequency

Heatmaps all at once is useful.

I did use the tables before, but would have preferred if it showed scores/letter rather than scores/key. Helps in the optimising process.

Suggestion for when you're bored: On Misc page, add "words on home block" where home block is 18 keys under 3 main fingers ( and thumbs?).


mstacker

  • Member
  • **
  • Posts: 27
    • View Profile
Re: Balanced Keyboard Layout
« Reply #998 on: October 17, 2017, 10:30:05 AM »
Gotta love Amazon: "This item does not ship to South Africa."
By the way. If you want, I can procure a few sets and ship them to you. Probably take a month. The real downside is, you'd likely have to take a pair of scissors and cut some of the stickers up as Dvorak keys are QWERTY rearranged. You'll discover that 'fun,' if you have not already, when you start to program the tiny for you dream keyboard.

Matt

iandoug

  • Hero Member
  • *****
  • Posts: 622
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #999 on: October 17, 2017, 11:17:23 AM »
By the way. If you want, I can procure a few sets and ship them to you. Probably take a month. The real downside is, you'd likely have to take a pair of scissors and cut some of the stickers up as Dvorak keys are QWERTY rearranged. You'll discover that 'fun,' if you have not already, when you start to program the tiny for you dream keyboard.

Thanks for the offer, I have family in UK that I use for such things :-)
Already used to the "cut up tiny stickers" procedure :-)

Cheers, Ian