Welcome, Guest. Please login or register.
Did you miss your activation email?
2020-Oct-20 07:01

Login with username, password and session length

Recent

Shoutbox

break:
May. 09 2020 - 10:02am
Awesome, happy to be back on the forums !
I hope everyone is in good health.
ADMIN:
May. 01 2020 - 6:54pm
new domain: ieants.cc
Den:
Mar. 20 2020 - 2:40am
possible delay of service
break:
Mar. 14 2020 - 1:05pm
@Den
Hey, had some issues opening the Worksheet in the new MS-Excel. Made a topic about it.
break:
Mar. 07 2020 - 12:19pm
@Den 
You are awesome, this helps a lot !  Really appreciate your help.
Den:
Mar. 07 2020 - 4:07am
.tbl file and translation program
break:
Mar. 05 2020 - 10:44pm
@Den Wow friend, you are blazing with these updates ! Question, how did you edit the names of things ?
Den:
Mar. 03 2020 - 1:21am

Author Topic: [BEAKL] Balanced Effortless Advanced Keyboard Layout  (Read 350615 times)

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: Went back to the drawing board ....
« Reply #1950 on: 2020-Sep-27 16:31 »
I'll just be lazy and wait for the plots. ;)

I have been thinking about those ... issues I have are the size (97 x 97 is a lot bigger than 27 x 27), and what sort of gradient ... eg is a 5% step okay or do we need a 1% step ... or 100/255 step.

This tool does it, but since most of the values are in low single digits or < 1, that detail gets lost ...

http://www.herethere.net/~samson/php/color_gradient/?cbegin=FFFFFF&cend=330033&steps=100


Mmmm might work better with 4 colours, each doing 25% in 1% steps.


moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: Went back to the drawing board ....
« Reply #1951 on: 2020-Sep-27 16:44 »
On the import dialogue, select Tab as separator, and make sure string delimiter is set to nothing.

That did the trick! Thanks!

I need to look in more detail when I have time, but the first impression is that for the first time explains why the brackets score so much better when they are on the hand that has 'hiea', including why putting the '(' on my 'i' key scores better than on my 'e', which never made sense for me.

btw) to display this data as bitmap you can always rescale to a 0-255 scale; but the key thing you really want to see is which ones are low and which ones are high and that is difficult to spot when just looking at numbers. The approach by thinqu where he/she split it into separate tables worked well for me (the ones under same finger movement here: https://microexploitation.com/2018/06/26/the-development-of-the-thinqu-keyboard-layout-factors-that-influence-typing-effort/)

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: Capacitive touch keyboard
« Reply #1952 on: 2020-Sep-30 02:52 »
So I have been thinking (which is always dangerous) about using "touch" keyboards.... the above designs clearly won't work because they're ANSI/ISO QWERTY and not what I want ...

Then it dawned on me that maybe I could use a touch-screen monitor instead, with a suitable on-screen keyboard.

Stumbled upon last night. It's only a matter of time before they replace the keyboard with a screen as well. Asus laptop with "screenpad".


iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1953 on: 2020-Sep-30 04:09 »
It might be interesting to see what layout ideas KLATest likes with these weightings (from my post above)... IIRC these values were based on strength AND flexibility of the fingers .. they're not mine, they come out of some formal research. The link should be somewhere in this thread.

And co-incidentally, my search for previous message re touch screens also turned up the post about finger strengths...

https://ieants.cc/smf/index.php?topic=89.msg2197#msg2197

and

https://ieants.cc/smf/index.php?topic=89.msg2198#msg2198

which have links to the papers (with curiously large counters... :-) )

Cheers, Ian

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1954 on: 2020-Oct-04 08:13 »
The story: I made a vertically staggered split mechanical keyboard Keebio Iris From a kit, I can customize it to my heart’s content.
I’m a programmer, I use vim.

I found BEAKL and I decided to learn touch typing. Finally.

I like the idea behind BEAKL, it’s very well thought out and precise, compared to most other alt layouts.
So, I started from 10 WPM today.
My BEAKL15 layout is not ready, but I can learn on keybd, all letters work.

lurk3r4sh

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1955 on: 2020-Oct-04 11:44 »
Story time!
I use this Nirvana-mirrored-back-ansi layout
I believe it's good enough and performs best on orthogonal ortholinear kb
4th day - 40 wpm

Previously have reached 90wpm on x1-ansi
My friend was sure Workman is better than Cole, and that's it.
I told her about x1.
She's instantly felt in love with the concept and is learning it now :)

Thank you, people.

P.S.
The nirvana layout listed on kb-design is a bit outdated.
I wasn't aware that site has any connection with the forum and started learning it :o.
« Last Edit: 2020-Oct-05 00:54 by lurk3r4sh »

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1956 on: 2020-Oct-04 13:28 »
The nirvana layout listed on kb-design is a bit outdated.
I wasn't aware that site has any connection with the forum and started learning it :o.

Um yes ... I should have changed the name to JCX or something but I like the Nirvana name....

I had to do a 3-way rotate to make it work on ANSI, don't think I've posted it yet. Due to the swaps it probably has a different feel to ortholinear version. I prefer ortholinear version.

I would swap C and F, but scores higher as attached.

I think the changes on ANSI are side-effect of shift being on pinkies.


« Last Edit: 2020-Oct-04 13:32 by iandoug »

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1957 on: 2020-Oct-04 13:54 »

I found BEAKL and I decided to learn touch typing. Finally.
My BEAKL15 layout is not ready, but I can learn on keybd, all letters work.

Some practice files for you. These ARE actual words, even if not very common ones. Week 1 is home row.

Cheers, Ian
« Last Edit: 2020-Oct-04 13:57 by iandoug »

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1958 on: 2020-Oct-04 13:59 »
Cool! I’ll try them!
I use vim since 1988 so, I’m quite used to it. But, always looking at the keys :(

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1959 on: 2020-Oct-04 14:43 »
What do you think about Tap-dance functionality? It’s from qmk firmware for keyboards and it means you can click a number to times to get different effect. It’s important when using smaller factors like 60% etc.
https://beta.docs.qmk.fm/using-qmk/software-features/feature_tap_dance
How to measure two clicks effort vs. one? Speed?

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1960 on: 2020-Oct-04 14:59 »
What do you think about Tap-dance functionality? It’s from qmk firmware for keyboards and it means you can click a number to times to get different effect. It’s important when using smaller factors like 60% etc.
https://beta.docs.qmk.fm/using-qmk/software-features/feature_tap_dance
How to measure two clicks effort vs. one? Speed?

I remember the topic coming up a while back ... KLA has no concept of such functionality... this actually touches on recent discussion about "keyboard layout definition" files .. in KLA we would need a way to define that behaviour in the Json, as well as code in the configuration and scoring to handle it.

KLA probably needs the concept of "auto-repeat" as well ... AFAIK at the moment it counts things like ######################### as multiple keypresses.

@Den.. was thinking in the shower (as usual) that there may be a case for "auto-switching" which thumb presses the space bar, instead of it being solely assigned to a single thumb ... ie autoselect thumb on opposite hand to last letter, except in cases of multiple spaces....

Cheers, Ian

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1961 on: 2020-Oct-04 15:11 »
I see - makes sense
I moved on a bit, made a layer with punctuations - this is the good stuff. I love the idea of . and @ on one key - emails!
Also, I use lots of {} to move in vi - very convenient. I got really interested in this, I will start on Monday to collect my own letter usage data, including vim commands. There is lots of w y c b $#*”’ etc etc. might be interesting corpus.
How many keys will be enough?

lurk3r4sh

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1962 on: 2020-Oct-05 02:13 »
Quote
I would swap C and F
Yeah. It would keep C in standart position.
Therefore require less mental effort, if you're used to C-c

But I'm not making that compromise and mirroring the layout back
+ having vowels on the left mb preffered (read it here, sorry that couldn't find now)

About K position.
Isn't it easier on ortholinear to have it on top row:
1 horizontal u away from the O
than on bottom:
2vu away?

And, If it is -- un-mirrored variant allows supperior top row K position on ANSI
« Last Edit: 2020-Oct-05 02:19 by lurk3r4sh »

lurk3r4sh

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1963 on: 2020-Oct-05 02:49 »
Quote
I had to do a 3-way rotate... don't think I've posted it yet
Interesting. I'd like to read more logs about nirvana creation and to know what is 3-way rotate :)

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1964 on: 2020-Oct-05 07:30 »
+ having vowels on the left mb preffered (read it here, sorry that couldn't find now)

That was an observation I made based on years of playing with layouts on KLA.

Since then I think I stumbled onto the reason .. it's because by default KLA ANSI layouts put the spacebar on the left thumb... and that setting largely got overlooked/ignored on ANSI.

But switching thumbs can make a dramatic difference. See results below where I just switched thumbs. This led to my recent comment above about auto-switching which thumb is used on ANSI.

So a better guideline would be "spacekey on same hand as vowels".

About K position.
Isn't it easier on ortholinear to have it on top row:
1 horizontal u away from the O
than on bottom:
2vu away?

And, If it is -- un-mirrored variant allows superior top row K position on ANSI

Yes I have it like that on Matrix. On ANSI with vowels on right, it's better on bottom.
Your mirrored version improves the K but you could maybe swap W and V.


iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1965 on: 2020-Oct-05 07:37 »
Interesting. I'd like to read more logs about nirvana creation and to know what is 3-way rotate :)

If you compare matrix above with ANSI below:

B -> H
H -> C
C-> F
F -> B

so actually 4-way not 3-way.

Due to ANSI stagger and C more frequent than F, scores better with C on top row.

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1966 on: 2020-Oct-05 07:39 »
How many keys will be enough?

more is better but KLA starts getting slow once input text goes above 60-80k range.

lurk3r4sh

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1967 on: 2020-Oct-05 12:41 »
Thanks.
I hope to have gergoplex someday, so I'd use latest nirvana-matrix available, mirrored version --- additional reason why:
I'd map KMonad/QMK layer switches on punctuation keys
Like that left hand could easily reach them (while right is doing other stuff) ---

Right now it's 20200924-test with V&W swapped (as you suggested)
« Last Edit: 2020-Oct-05 12:46 by lurk3r4sh »

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1968 on: 2020-Oct-06 15:34 »
So a better guideline would be "spacekey on same hand as vowels".

True, but it is also worth keeping in mind that the percentage of people that is right-handed is larger. This in itself makes putting the vowels on the left-hand preferential as that way you have the least able/flexible hand make less movements if one optimizes for alternation. Obviously the fact that an ANSI keyboard also has more keys on the right-hand pushes the balance even more in that direction as you need more keys to cover the non-vowels.

The real nuisance here is the interaction between common keyboard short-cuts and working with a mouse; not an issue when using a thinkpad though :-)

btw) I don't think is included in the model though, so still think there is something weird in the KLA modeler scoring perfectly mirrored layouts differently

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1969 on: 2020-Oct-06 15:41 »
Right now it's 20200924-test with V&W swapped (as you suggested)

Looks familiar. Might be worth comparing it against the one I posted here: https://ieants.cc/smf/index.php?topic=89.msg3069#msg3069 (that one has the highest score I could get for that home-row).

Note that although the layout is displayed as ergolinear I also use my layouts on my laptop, with the tweak of moving the home-row position of the right-hand one key to the right.

btw) I've moved back to putting the 'R' on pinky since then. My current layout is in later posts....mainly a case of finding less movement more comfortable at the expense of slightly higher load on the pinkies.
« Last Edit: 2020-Oct-06 15:47 by moesasji »

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1970 on: 2020-Oct-06 16:33 »
BEAKL seems easy to learn. I tried to work normally with my iris today - it was hard slow and exhausting. But, very satisfying.

I noticed that vim major commands are doable. Change word cw is weird, but I can map it to unused cc in vim normal mode.

I tried to log keystrokes, there is small problem, as vim logs either one byte or three bytes.
Good read here: https://vi.stackexchange.com/questions/14441/how-does-vimgolf-record-all-keystrokes

I can convert them to anything.

The question is - what format do you need for corpus?

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1971 on: 2020-Oct-06 18:43 »
The question is - what format do you need for corpus?

KLE tries to map the input character to what is on the keyboard, and ignores anything it can't find.

It only understands a few non-character keys. It kinda expects ASCII I think ... Den will know better. I think it can handle diacritic characters but things like ctrl-C, I dunno... it's never come up. Den?

That didn't help much :-(

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1972 on: 2020-Oct-07 15:44 »
OK, I log, but, it will not be helpful. Lots of JK etc.
Second day. Very hard but somehow a bit faster. I still improve layout ctrls alts etc. I put minus and underscore on left control and so on.

What is best way to reach further keys, like Z - move hand with index on S? When I just move hand I have problem with next key sometimes. Practice!!

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1973 on: 2020-Oct-07 15:47 »
Another question - do you have a chart comparing metrics with Dvorak, Colemak etc for BEAKL15, similar to the one from first page of this thread?
Found: https://ieants.cc/code/keyboard/klatest/#/main
« Last Edit: 2020-Oct-07 16:39 by Tomdraug »

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1974 on: 2020-Oct-07 19:41 »
Revisited an approach from some time ago in which I use the number-row on a standard European layout. It is shocking how well it performs in klatest, which I had forgotten about. The gap is big enough that I'm actually wondering whether I am fooling the optimizer somehow.

@Ian: any thoughts why this one is doing so well compared to the same style of layout on Ergolinear?


iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1975 on: 2020-Oct-08 02:40 »
Revisited an approach from some time ago in which I use the number-row on a standard European layout. It is shocking how well it performs in klatest, which I had forgotten about. The gap is big enough that I'm actually wondering whether I am fooling the optimizer somehow.

@Ian: any thoughts why this one is doing so well compared to the same style of layout on Ergolinear?

It reminds me of Arensito in some ways.

My checker program says your Json contains errors and can't be checked.

But I can't immediately see it, and it loads in KLA so will need to read it carefully...

Cheers. Ian

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1976 on: 2020-Oct-08 03:19 »
My checker program says your Json contains errors and can't be checked.
But I can't immediately see it, and it loads in KLA so will need to read it carefully...

File has Byte Order Mark flag set. Checks okay.

Layout does well for 4 reasons:

1. actual letter layout is already proven good.
2. low pinky usage, which KLATest likes
3. finger allocation, which relates to (2) and home position shift
4. You took some punctuation off AltGr which reduces finger usage.

Cheers, Ian

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1977 on: 2020-Oct-08 09:55 »
Layout does well for 4 reasons:

1. actual letter layout is already proven good.
2. low pinky usage, which KLATest likes
3. finger allocation, which relates to (2) and home position shift
4. You took some punctuation off AltGr which reduces finger usage.

I still wouldn't expect it to beat know good layouts like the BEAKL Opted4 Ergo Alt by such a big margin (8-10% in many) and easily beating my previous bests by the same margins. I need to put this on the same ergodox template as the Beakl Opted4 to make it a fairer comparison.

btw) the thinking behind removing the punctuation from the AltGr layer is that way it gets a lower frequency than the shift, so I can put the shift on home.  I hadn't spotted that the last time I played with the number-row.

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1978 on: 2020-Oct-08 10:40 »
I still wouldn't expect it to beat know good layouts like the BEAKL Opted4 Ergo Alt by such a big margin (8-10% in many) and easily beating my previous bests by the same margins. I need to put this on the same ergodox template as the Beakl Opted4 to make it a fairer comparison.

minor tweaks should get minor improvement


moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1979 on: 2020-Oct-08 14:25 »
minor tweaks should get minor improvement

Those really depend on the text you feed it as this does worse on academic type text, but I agree it can be tweaked to score higher. Moving the - and _ to the position of the j also tends to help.

The key thing that I would really like to fix first is the get cursor movements to a better position and don't yet see how to best fit them into this layout without loosing the ease with which I can type numbers . By now I'm pretty sure that the cursor-movement should be on the opposite hand to the shift/altgr and on the home-row, but I haven't yet found a good solution for that. 

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1980 on: 2020-Oct-08 14:43 »
best fit them into this layout without loosing the ease with which I can type numbers .

I use two layouts and altgr to accomplish that. I have raise layer and lower layer on top of BEAKL 15. Lower is with puncs and raise is with arrows on the right home row, numbers around left home and other navs everywhere else. I even have ctrl-alt-del as separate key there (sadly, I must use windoze at work)

I have 6 thumb keys: ctrl/- enter/lower space/raise under left thumb and esc/shift tab/altgr comma/alt under right. I use mod tab to have double action on all my thumbs which was quite crazy at the beginning.

I use altgr solely for Polish special letters.

I noticed that two layouts, which are not too packed up are easy to type and remember.

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1981 on: 2020-Oct-08 18:26 »
I use two layouts and altgr to accomplish that. I have raise layer and lower layer on top of BEAKL 15. Lower is with puncs and raise is with arrows on the right home row, numbers around left home and other navs everywhere else. I even have ctrl-alt-del as separate key there (sadly, I must use windoze at work)

Over time I came to the conclusion that I prefer the combination of just two layers in combination with sticky modifiers and sticky keys to switch the layer. Mainly because I do need to be able to use my layout on a laptop, which is more restricted in terms of "useful" thumb-buttons assuming you can't get one with a Japanese layout.

minor tweaks should get minor improvement

Done a bit more tweaking while putting my cursor-movements on relatively logical keys on the same layer. The most used are up and down and are the same as one would use in both vim and emacs in combination with CTRL, so easy to remember. The attached version does astonishingly well in klatest on classics; academic text and coding with the screenshots showing a pretty clear gap.  :)

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1982 on: 2020-Oct-08 18:44 »
Excellent results!
Good idea with stickies I must try it.

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1983 on: 2020-Oct-10 06:51 »
Excellent results!

Thanks, but this was not yet the optimum this layout can do. Putting the Enter on a more optimal position at the expense of putting - and _ back onto the number row removes another 0.5 to 1 from the total score and  seems pretty optimal to me even though I'm not a 100% happy with the reach to type the - and _.

btw) Note that on Ergodox I would put the backspace and Tab on the thumb (and depending on your favorite editor one can swap the Tab with the Escape)
« Last Edit: 2020-Oct-10 07:53 by moesasji »

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1984 on: 2020-Oct-10 11:14 »
In vim i have dispropportionate number of wzdgrcp commands especially w and p which I use a lot. Also I use r often as in c-r for history in bash. So, it poses interesting question. BEAKL 10 is better in this regard.

How about vimak layout?  ::)
« Last Edit: 2020-Oct-10 11:19 by Tomdraug »

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1985 on: 2020-Oct-10 11:16 »
on Ergodox I would put the backspace and Tab on the thumb (and depending on your favorite editor one can swap the Tab with the Escape)
Yes, I have now esc backspace enter and space on thumbs. They are very surprised, so much work!

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1986 on: 2020-Oct-10 15:03 »
In vim i have dispropportionate number of wzdgrcp commands especially w and p which I use a lot. Also I use r often as in c-r for history in bash. So, it poses interesting question. BEAKL 10 is better in this regard.

I have used Vim for a very long time and more recently been playing with emacs and kakoune. BEAKL10 has many similarities in the letter placement, so I don't really see why it would be substantially better for Vim.

Note that one of the key tricks I use that makes it score so well is that by moving common symbols to the main layer by using that the number-row is actually easy to reach for middle and ring finger I can put the shift under the thumb. This is because putting less symbols of on the AltGr layer dramatically lowers the need for the AltGr. In practice having the shift under the thumb makes routine typing and using Vim a lot more pleasant as capitals are actually pretty common in the vim commands.

Direct comparison on klatest below that includes BEAKL10 for classics, one of the academic texts and the hanoi coding one below. If we believe the optimizer the numbers should give some food for thought:

edit: for clarity - my bl13-iso layout for some reason does particularly well on klatest.....on Den-v2 it doesn't score this well!
« Last Edit: 2020-Oct-10 15:25 by moesasji »

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1987 on: 2020-Oct-10 15:46 »
Thank you for exhaustive answer.
Shift under thumb is a game changer.
Anyway, I learned 15 so I will work it for a while.
Btw I put numbers in left home box on separate layer and I no more need top row at all. It’s hard to type there :)
My next keyboard will be 3x6 and 4 thumbs per side

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1988 on: 2020-Oct-11 03:13 »
I need to put this on the same ergodox template as the Beakl Opted4 to make it a fairer comparison.

Good news. If I put the same basic layout on an ergodox as a sanity check the scoring improves further as one would expect. So this layout doesn't do well just because there is something weird going on because I was using a standard European layout. It genuinely does well under the scoring rules for klatest. Surprising but true!

To play the json files and screenshots showing how it compares for classics, academic and coding below. Note that I still find the massive lead it takes over the Beakl4 Opted Ergo astonishing!

Tomdraug

  • Member
  • **
  • Posts: 15
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1989 on: 2020-Oct-11 06:03 »
I run the tests and I get consistent great results.
I run also on Polish 62k text, result below
« Last Edit: 2020-Oct-11 06:10 by Tomdraug »

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1990 on: 2020-Oct-11 06:49 »
minor tweaks should get minor improvement

To follow Ian's advice....attached some tweaks to make it slightly less annoying for Vim and improve the AltGr layer for coding. This seems very close to the best I can get with this home-row.

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Excessive use of a finger
« Reply #1991 on: 2020-Oct-11 17:06 »
@Den

Just wondering if there is a case for adding a penalty for "excessive use of one (or more) fingers, or if this is already adequately catered for in the scoring.

Since we're aiming for "balance".

As always, how much is too much, and what should the penalty be...?

For example, just did a test where my X7.1 layout had 19 + 9% presses on the two thumbs ... ie 28% of all keystrokes were on the 2 thumbs. Perhaps this is excessive?

I open the topic to the floor for discussion... :-)

Cheers, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1992 on: 2020-Oct-12 06:43 »
To follow Ian's advice....attached some tweaks to make it slightly less annoying for Vim and improve the AltGr layer for coding. This seems very close to the best I can get with this home-row.

You need to put AltGr on the right thumb, unless you have a particularly interesting index finger.... :-)

Cheers, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Here we go again...
« Reply #1993 on: 2020-Oct-12 07:02 »
Given recent efforts by others, it became necessary to optimise more ...

Attached two layouts which do well at most things English. They struggle a bit if there are a lot of numbers, but I would probably switch to numpad for those in real world.

I would use the simpler version myself, because I put a lot of other non-ANSI stuff on AltGr. The AltGr version is to compete more directly with other new layouts, especially for code.

@Den: there seems to be an annoying bug of late ... when you load KLEtest, import a layout, and edit a key, it seems to reload and lose what you've done (including loading input text). After that it's okay, except for the growing memory usage and gradual slowdown. I've tried Firefox, Falkon (Chrome) and Vivaldi (Chrome).

@Den2: can you maybe add attached to the Code section ... it is the "new" code corpus I was compiling before I changed direction and pulled in 99% of RossettaCode.
This one file has samples from multiple popular and some legacy languages, plus KLE (for HTML), all in one file, and thus provides a more "comprehensive" code test.

I may still fiddle with the non-AltGr a bit ... seems putting some brackets with vowels is better.

Thanks, Ian

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1994 on: 2020-Oct-12 07:07 »
You need to put AltGr on the right thumb, unless you have a particularly interesting index finger.... :-)

Well spotted. That is a translation mistake when moving it from the European to the Ergodox.....weirdly enough it actually lowers the score a bit.

btw) in practice I'm using the heiatnsr variant of this; it scores not as good, but I think it is the better one as the long jump from R to L and vice versa really annoys me in practice on hieatnsc
« Last Edit: 2020-Oct-12 07:10 by moesasji »

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: Here we go again...
« Reply #1995 on: 2020-Oct-12 13:56 »
@Den2: can you maybe add attached to the Code section ... it is the "new" code corpus I was compiling before I changed direction and pulled in 99% of RossettaCode.
This one file has samples from multiple popular and some legacy languages, plus KLE (for HTML), all in one file, and thus provides a more "comprehensive" code test.

@Den: Can you maybe add attached to English texts ... I received permission from the author via email today.

The letter frequency for the first 15 letters more closely matches accepted English frequency than any of the other texts.

Also, is it possible to add a date-stamped note if/when you tweak the scoring? Just so that I don't think I'm going mad :-)

Thanks, Ian

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: Here we go again...
« Reply #1996 on: 2020-Oct-12 15:01 »
Given recent efforts by others, it became necessary to optimise more ...

Attached two layouts which do well at most things English. They struggle a bit if there are a lot of numbers, but I would probably switch to numpad for those in real world.

I would use the simpler version myself, because I put a lot of other non-ANSI stuff on AltGr. The AltGr version is to compete more directly with other new layouts, especially for code.

Great to see the competition lead to testing out new ideas as this looks impressive!!

Even though I'm almost willing to admit defeat, this scoring really makes me think we are hitting weird behavior in KLA.

To demonstrate why I think this: If I just mirror your layout (and fix the orientation of the brackets to match) it scores worse than mine that has the mirrored layout. To show this: below the files of the mirrored nirvana.AltGr version compared against the one you posted using the classics collection. That gives a gap in of 7% in the layouts that based on a matrix being a symmetric keyboard should really score identical. This behavior really smells like a bug to me!

The behavior that looks wrong is that the mirrored versions score very differently on the same hand metric, with all the other numbers the same as one would expect. There is really no logic why that should be the case....a case of an array being of by one somewhere in the code?

iandoug

  • Hero Member
  • *****
  • Posts: 1115
    • View Profile
    • Keyboard Design
Re: Here we go again...
« Reply #1997 on: 2020-Oct-12 16:56 »
The behavior that looks wrong is that the mirrored versions score very differently on the same hand metric, with all the other numbers the same as one would expect. There is really no logic why that should be the case....a case of an array being of by one somewhere in the code?

I wonder if this is related to the very different scores you get on ANSI when swapping which thumb hits the space bar. I thought it odd that it would so dramatically affect layouts with vowels and consonants all over the place.

Den may know.

Cheers, Ian

moesasji

  • Valued Member
  • ***
  • Posts: 84
    • View Profile
Re: Here we go again...
« Reply #1998 on: 2020-Oct-13 07:30 »
I wonder if this is related to the very different scores you get on ANSI when swapping which thumb hits the space bar. I thought it odd that it would so dramatically affect layouts with vowels and consonants all over the place.

The reason I tried this sanity check is indeed that I had stumbled upon the above issue before, but now the case is clearer as the matrix layout is by definition a symmetric one. The fact that just mirroring the layout gives a different score for the words one can type with the same hand smells like a bug to me. 

sdothum

  • Member
  • **
  • Posts: 48
    • View Profile
Re: [BEAKL] Balanced Effortless Advanced Keyboard Layout
« Reply #1999 on: 2020-Oct-13 11:21 »
Just a pointer to a relatively new python based keyboard analyser which can be found here..

https://github.com/bclnr/kb-layout-evaluation

The current codebase does not handle the thumb row (and, hence, space key) and layers, so it is somewhat basic. But the config file is quite readable and allows defining a finger roll penalty scheme. The delault weighting scheme is based on the ColemakDH dev page so favours the more traditional QWERTY layout alternatives, with BEAKL in the running but downstream from MTGAP. Basically, any layout is better than QWERTY and its variants (unsuprisingly), the differences numerically not being that far apart.

Plugging in BEAKL weightings, of course, pushes BEAKL layouts to the top. The code is in python so is easily modifiable for the interested (I added a thumb row to see how my "thumb I" BEAKL variant would fair against BEAKL 15 and it holds its own).

Anyway, thought some of you might be interested in this code base since you could extend it to accommodate the rules and evaluation schemes of import to you (eliminating the ambiguity of how other analysers work).

I haven't taken a deep dive into the code to determine how the corpus is defined and processed (but it appears to use a pre-processed bigram count file vs processing a sample text file.. so a parser would be in order to run random files against this program).

« Last Edit: 2020-Oct-13 13:23 by sdothum »

 

2 Guests, 0 Users