Welcome, Guest. Please login or register.
Did you miss your activation email?
2019-Jun-24 09:29

Login with username, password and session length

Recent

Shoutbox

break:
Mar. 24 2019 - 7:46pm
@Den Just finished 2nd playthrough and testing session. I really like the edits I have made. Going to take another week to look over things !!
break:
Mar. 11 2019 - 7:04pm
@Den Hey, thanks for all of the support. Finished with the changes and currently playing through the game !!
break:
Feb. 25 2019 - 7:31pm
@Den Yo, finishing up the bosses and will do some more testing before the big release !!
break:
Feb. 18 2019 - 10:42pm
@Den Hey, got the main bosses edited and working on the treasure chests. Getting close to finished my edits !!
break:
Feb. 10 2019 - 9:11pm
@Den Yo, got the shops edited now and will start work on the treasure edits. I bit confused on monster chests ATM...
break:
Jan. 29 2019 - 6:56pm
@Den Hey, really making progress with my Hex-Editing. Replied to my topic...a bit stuck again
break:
Jan. 23 2019 - 11:42pm
@Den  Yo, I am really getting the hang of this HEX editing stuff !! Working on figuring out the growth rate formula !!
break:
Jan. 17 2019 - 8:31pm
@Den I am starting get more understanding about Hex editing.  Thanks for taking time out for me !

Author Topic: Balanced Keyboard Layout  (Read 275145 times)

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1500 on: 2019-Jan-14 04:07 »
So I came back after a while and see that beakl 15 has replaced beakl 9 as the recommended layout. It seems like it is very similar to beakl 9, and most of the changes are to a few character in the marginal positions on the left hand. Can I ask anyone who has used both whether they think the switch is worth it?

The benefits of 15 over 9 are:
 less movement to far index column
 less same finger
 better rolls
 better bigrams and trigrams
Support me on Patreon

I saw. I conquered. I came.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
It's a keyboard, Jim, but not as we know it
« Reply #1501 on: 2019-Feb-05 10:13 »
Lifted from Massdrop.

Discuss.

Not sure what to make of the design or what the designer was trying to achieve.

The two Bs are interesting.

Guess the arrows and other essentials are hiding there somewhere.

Would not suit my typing style.

shopt

  • Newbie
  • *
  • Posts: 9
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1502 on: 2019-Feb-06 00:03 »
That still looks like it has the standard "designed for 2 right hands" problem that most keyboards do.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: It's a keyboard, Jim, but not as we know it
« Reply #1503 on: 2019-Feb-07 00:28 »
Pinky keys farther from other fingers, stretching pinky to even more uncomfortable degree.

Is doubling up B necessary? The more contentious key is number 6. Controversial among typists which hand should hit 6. So if any key should be doubled, it should be 6.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: It's a keyboard, Jim, but not as we know it
« Reply #1504 on: 2019-Feb-07 01:20 »
Pinky keys farther from other fingers, stretching pinky to even more uncomfortable degree.

Yeah, I think some of those angles are a bit odd. Looks like they slightly lowered pinky home, but changed the angle as well and I'm not sure how comfortable that would be.

Is doubling up B necessary? The more contentious key is number 6. Controversial among typists which hand should hit 6. So if any key should be doubled, it should be 6.

If you look at my current (still!!!! ... must sort out software on new one) keyboard, MS Natural, "N" key, it's got this awkward dead space. So I guess they thought they should fill it with something for aesthetic reasons, touch-typing skills be damned.
Pic from Wikipedia:

https://en.wikipedia.org/wiki/Microsoft_ergonomic_keyboards#/media/File:MicrosoftNaturalKeyboardGen1.jpg

I also looked at the 5/6 split and wondered about it :-)

Cheers, Ian
« Last Edit: 2019-Feb-07 01:26 by iandoug »

qites

  • Newbie
  • *
  • Posts: 2
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1505 on: 2019-Feb-09 03:38 »
Hello everyone again!
This time I adapted the BEAKL15 to the laptop keyboard.
Also here is the Cyrillic phonetic layout
And it works well on both ISO / US keyboards.



I have been living with this for half a year and no longer dream of a real ergonomic keyboard!

You are the best!

Thanks to GoogleTranslate for translating this message.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1506 on: 2019-Feb-10 12:38 »
Glad BEAKL is working out for you even on unergonomic keyboard.

It's funny though once you condense all the keys around home block, that leaves a lot of empty unused keys on the outskirts.

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1507 on: 2019-Feb-10 16:15 »
Hi everybody, back after a while.

I had been using BEAKL15 on and off in the past and just as I thought of swapping G - W (G on qwerty Y is a tough reach on my Microsoft Ergo Sculpt !) .. I saw BEAKL16 on Den's KLA, which does swap G-W already.
It seems to be having good scores, usually somewhat better than 15, any reasons why BEAKL16 is not the new recommended layout vs BEAKL16 ??

The new BEAKL PLLT x1 is pretty interesting, but is really not too good when I try typing french ;-) 
ps: I use spacebar to access the 2nd layer, so having Space in the main cluster works great.

Thx

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1508 on: 2019-Feb-10 18:23 »
BEAKL 16 attempts to emphasize more load for the stronger right hand. unfortunately it retains too much effort on the index inner column. and H at top ring may not feel good as outward roll.

in your case, try swapping the G and a punctuation from the left hand. so G ends up on left hand. and the far top index key for a rare punctuation.

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1509 on: 2019-Feb-10 18:53 »
BEAKL 16 attempts to emphasize more load for the stronger right hand. unfortunately it retains too much effort on the index inner column. and H at top ring may not feel good as outward roll.

in your case, try swapping the G and a punctuation from the left hand. so G ends up on left hand. and the far top index key for a rare punctuation.
Oh ok, I'll look at 15 again, I'll check out your suggestion.

Here's the other layout I was using sometimes
(self made with modified MTGAP, using some BEAKL principles adapted to perso tastes and US std kbd)
Code: [Select]
  q  p  o  u  "   '  n  d  f  v
; g  i  a  e  x   l  r  t  s  c `
  k  ,  .  y  z   j  h  w  m  b

  Q  P  O  U  "   '  N  D  F  V
  G  I  A  E  X   L  R  T  S  C  ^
  K  /  ?  Y  Z   J  H  W  M  B
' " ^ used as dead keys for French

thx again, I'll let you know ;-_)

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1510 on: 2019-Feb-10 21:05 »
Here's the other layout I was using sometimes

Interesting use of CapsLock key :-)

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1511 on: 2019-Feb-10 22:10 »
Interesting use of CapsLock key :-)
Hehe, I do this also with BEAKL1?, so I have ` on the right, for alternation when doing à etc

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1512 on: 2019-Feb-10 22:27 »
in your case, try swapping the G and a punctuation from the left hand. so G ends up on left hand. and the far top index key for a rare punctuation.
Hmm, did not do so well, "same hand" went up (ok not much, but still).
Swapping G-W gave better results (Den's KLA), actually about +2% better than std BEAKL15 on the french text, same score for Classics.
I'll give this a try :)
(oh, I also swap , / and move ; to CapsLock, I use Shift-CapsLock for CapsLock)
Code: [Select]
  4 0 1 2 3            7 6 5 9 8   
  q h o u x            w c r f z   
; y i e a .            d s t n b `
  j , / k '            g m l p v   

  Q H O U X            W C R F Z   
  Y I E A @            D S T N B _
  J ! ? K ^            G M L P V   
 
« Last Edit: 2019-Feb-10 22:44 by philippe.quesnel »

AZERTYSucks

  • Member
  • **
  • Posts: 15
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1513 on: 2019-Feb-13 15:14 »
Hey Den, do you think you can re-upload the .cfg file you used for opt? The link on your site seems to be broken.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1514 on: 2019-Feb-14 01:51 »
Hey Den, do you think you can re-upload the .cfg file you used for opt? The link on your site seems to be broken.

Navigate to the parent directory to find bunch of cfg files. Most are experimental. But one of the released layouts is beakl8.cfg

Also the variant beakl8a.cfg gives this layout:

Code: [Select]
qyoix gcrfz
kheau dstnb
j/.,' wmlpv
« Last Edit: 2019-Feb-14 02:52 by Den »

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1515 on: 2019-Feb-15 03:39 »
KLA update:
; added corpus per Ian's recommendations
; added layout BEAKL19

Code: [Select]
BEAKL19

q.ouj wdnm,
haeik gsrtp
z'/yx bclfv

excellent low same finger and rolls

speaking of low same finger, need to revisit BEAKL Opted4 (and Arensito). which have ridiculous low same finger on current KLAtest configuration.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1516 on: 2019-Feb-15 03:51 »
; added layout BEAKL19

Mmm are the missing BEAKLs in some CIA black site? :-)

Please to post .json

Might revert on corpus, need to check current collection vs what is on KLA now.

Not sure what to do about all the missing letters which are typically not used in English... but need an optimal spot on the keyboard.
And using source code is too biased depending on programming language (or even environment like EMACS or vi ... but suppose they may qualify as edge cases as most people use Windows and thus won't use EMACS or vi.

Thanks, Ian

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1517 on: 2019-Feb-18 22:32 »
KLA update:
; added corpus per Ian's recommendations
; added layout BEAKL19

excellent low same finger and rolls

Hmm, interesting, I had noticed the new corpus today, but did not realize about BEAKL19 !! must check it out !
yes, still on the QUEST .. I keep going back t0 BEAKL15 (well my modified version, for US Kbd and French), that and one of my own MTGAP / BEAKL creation.
Hey I should try to create a layout with these letters 'QUEST' .. would be such a great name hahaha ;-)

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1518 on: 2019-Feb-18 22:33 »
Hi Den,

Do you use the -DOHNE2SHIFT option when building AdNW opt,
the one that "causes upper case letters that occur as the second character in a sequence of two characters to be ignored" ??

Oh and do you run -2 or -3 ?

merci
« Last Edit: 2019-Feb-19 01:21 by philippe.quesnel »

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1519 on: 2019-Feb-20 22:51 »
So I tried this (funny) idea of Q.U.E.ST layout:
fixed positions for Q/U/E/S/T
Did it with AdNW opt, BEAKL19 cfg applied to ANSI / US kbd,
(I did not bother to try to fix parameters that might not make sense when moved to ANSI kbd (?))
Code: [Select]
qhioz vdrlb   Q.... .....
guae' fsntp   .U.E. .S.T.
jy,./ xcwmk   ..... .....
Surprisingly, not as bad results as I would have expected (having fun after all) !!
In Den KLA / Alice In Wonderland and others, has almost identical results as 'ANSI version' of BEAKL15  :)
(actually a bit better in French ! haha, kind of surprised)

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1520 on: 2019-Feb-21 00:10 »
So I tried this (funny) idea of Q.U.E.ST layout:

How is same-finger usage on right index finger?

Thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1521 on: 2019-Feb-21 05:33 »
@Den

KLA update:
; added corpus per Ian's recommendations
; added layout BEAKL19

FWIW, your BEAKL PLLT x1 is not valid for KLA ... two keys needed at same time on right thumb. Unless you tweaked the code to allow that. By default it will ignore any letters needing that, leading to lower (better) score.

BEAKL Opted4 Ergo Alt continues to impress ... :-) ... despite its age.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Here be the X8
« Reply #1522 on: 2019-Feb-21 06:01 »
Minor tweak to X7.1H produces decent results on KLAtest.

Should also be more mainstream without the hH on the space key.

Move inspired by BEAKL Opted4 Ergo Alt.

I think we've probably explored that homekey layout a lot now. Clearly has something going for it.

Edit: discovered two backticks and no single quote, so removed Json, fixing, also tweaking to do better at code. Don't think I ever got around to optimising the X7 for code.
« Last Edit: 2019-Feb-21 09:07 by iandoug »

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1523 on: 2019-Feb-21 08:02 »
How is same-finger usage on right index finger?

Thanks, Ian
I'm on my phone right now, can check when I get to a computer .
Here's the json file of you want to check it out yourself

https://github.com/phques/txt2autokey/blob/master/examples.layers/AdNW/stdUs33Quest/kla/usKbd/AdNWansiQuest2AzAnsi.json

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1524 on: 2019-Feb-21 12:05 »
I'm on my phone right now, can check when I get to a computer .
Here's the json file of you want to check it out yourself

https://github.com/phques/txt2autokey/blob/master/examples.layers/AdNW/stdUs33Quest/kla/usKbd/AdNWansiQuest2AzAnsi.json
same finger for right index looks ok .. but left is pretty bad !!

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Here be the X8
« Reply #1525 on: 2019-Feb-21 12:19 »
Minor tweak to X7.1H produces decent results on KLAtest.

Should also be more mainstream without the hH on the space key.

Move inspired by BEAKL Opted4 Ergo Alt.

I think we've probably explored that homekey layout a lot now. Clearly has something going for it.

Edit: discovered two backticks and no single quote, so removed Json, fixing, also tweaking to do better at code. Don't think I ever got around to optimising the X7 for code.
is this still at https://kla.keyboard-design.com/#/config ??
thx

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Here be the X8
« Reply #1526 on: 2019-Feb-21 12:48 »
is this still at https://kla.keyboard-design.com/#/config ??

No, I'm playing on Den's Klatest site

http://shenafu.com/code/keyboard/klatest

Still fiddling, testing against code is tricky so kinda going in circles a bit.

My site seems a bit broken.... will investigate. Even local (on this PC) seems non-functioning ... probably some back-end script thing.
« Last Edit: 2019-Feb-21 13:02 by iandoug »

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Here be the X8
« Reply #1527 on: 2019-Feb-21 14:58 »
No, I'm playing on Den's Klatest site

http://shenafu.com/code/keyboard/klatest

Still fiddling, testing against code is tricky so kinda going in circles a bit.

My site seems a bit broken.... will investigate. Even local (on this PC) seems non-functioning ... probably some back-end script thing.
oh ok, thx.
I guess  it is not uploaded yet, I still an see H on space hehe ;-) (and you did say "still fiddling"! )
Yes, trying to get a layout to do good with code is a bit crazy, as you already mentioned, it depends so much on the language and also I guess the coding style etc !
Good luck with your site.
curious to see what the X7 changes are, that H on space thing did not work very well on ANSI / US std kbd (even if I DO use a type of "dual mode" on space and Shift_L for my layouts) and I rememner that it usually was one of the layouts that had good scores, that I had given a look at ;-)

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Here be the X8
« Reply #1528 on: 2019-Feb-21 16:20 »
curious to see what the X7 changes are, that H on space thing did not work very well on ANSI / US std kbd (even if I DO use a type of "dual mode" on space and Shift_L for my layouts) and I rememner that it usually was one of the layouts that had good scores, that I had given a look at ;-)

Attached more or less where I am... basically put hH back on left pinky. I've also swapped P and J, because "pi" as a bigram crops up too much in some of the English and code tests. (eg typing / typist, pick, etc.) Though suppose other texts will have lots of happy papa to confound things.

Struggling with the punctuation characters... the usual conflict between demands of HTML and JS vs demands of English.

The code tests in the corpus are not very good... think we will have to bite the bullet and go for selections of mainstream code, eg a file with various C, another with various HTML, another with various Ruby, etc.... but even then you still have the Tarzan/James Bond problem of whatever variable names the coder picked, influencing the letter frequency.


philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1529 on: 2019-Feb-21 23:10 »
interesting layout Ian.
Your comments and earlier ones made me realize that I was doing a mistake when comparing some layouts on KLA, I was doing quick tests without the AltGr punc layer, to save time (building the json) .. which gave falsely better results, in some cases.
I noticed it when I tried an 'ANSI / US' kbd version of X8.2 temp and realized there were no ',. etc
Doh ! Will have to revisit all those layouts I thought were doing good ;-p
This does explain why some of the "possibly fake good ones" did not always feel that nice to me when I tried them myself !

BEAKL19 / BEAKL15 were often on the top of the list, they probably will be THE top of the list when I redo the tests properly

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1530 on: 2019-Feb-22 01:15 »
Your comments and earlier ones made me realize that I was doing a mistake when comparing some layouts on KLA, I was doing quick tests without the AltGr punc layer, to save time (building the json) .. which gave falsely better results, in some cases.
I noticed it when I tried an 'ANSI / US' kbd version of X8.2 temp and realized there were no ',. etc
Doh ! Will have to revisit all those layouts I thought were doing good ;-p
This does explain why some of the "possibly fake good ones" did not always feel that nice to me when I tried them myself !

Random observations:

1. Trying to put an Alt-Gr style layout on ANSI/ISO physical keyboard may lead to finger damage ... the form factor is already wrong in so many ways, and adding the Alt-Gr gymnastics won't help. I think even Arensito's inventor uses a Kinesis/Ergodox. I know these boards are expensive and justifying the cost is hard... if necessary save $20 a month or so until you have enough, and don't tell Significant Other about the price.... :-)

2. My versions of KLA included an "ANSI chars" test ... basically a string of all the characters on an US ANSI keyboard. You run your design against this, and check the browser log for "character not found" errors. On Firefox this is via "webconsole" or other similar "developer" tool. Chrome also has this somewhere.
Here's the string of 95 chars: (space is second char for posting here, normally first char. Arrangement is in decreasing frequency.

e taoinsrhldcumpfg.ywb-,v0k1TAIS2C'"/3ED9:MN=RP;4OB5)L(HFx8W67G_UjqzJ<?Y@*VK!|$~[]%X&+#QZ}{>`\^

I have a PHP program that  I run to do bulk checking of all layouts so that I don't compare incomplete layouts and get misleading scores.
I should pretty it up a bit for posting. Current problem with it is that Den's version of ErgoLinear has two more keys than mine, so the keycount check fails. Need a workaround for that.

Cheers, Ian


From above string we can conclude that people smile ":-)" more than they frown ":-(".



iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1531 on: 2019-Feb-22 05:59 »
I have a PHP program that  I run to do bulk checking of all layouts so that I don't compare incomplete layouts and get misleading scores.
I should pretty it up a bit for posting. Current problem with it is that Den's version of ErgoLinear has two more keys than mine, so the keycount check fails. Need a workaround for that.

The program also loads analysis of the meta-data about the layout into a database, so it's a bit more complex than just a layout checker. Apart from the ANSI (and others for ISO) character checks, it also checks if all the other keys necessary for KLA are present, no duplicate chars, correct number of keys in layout, etc.

@Den: I see I specifically ignored Opted4 Ergo Alt, but can't see anything wrong with version currently on KLAtest.

I would only have ignored it if it was broken in some way and "fixing" it was not viable without author input. Currently ignore these:

$ignore = array (
                 "BEAKL Opted4 Ergo Alt", 
                 "Ergodox/Colemak/Cub",
                 "Ergodox/QGMLWB/Cub",
                 "Ergodox/QGMLWY/Cub",
                 "Ergodox/Workman/Cub",
                 "Ergodox Gelatin",
                 "TaserFR"
                );

Did you make changes between older version and current version on KLAtest?

thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
English corpus analysis
« Reply #1532 on: 2019-Feb-22 08:57 »
Attached spreadsheet has current state of corpus (English prose) analysis.
These are the "best" inputs, you will find no Alice or even most of the others that were in previous public KLA sites, either mine or Den's.

The file has fields separated by the tab character, opening this with something like LibreOffice calc is easy, don't know about Excel. Had to zip it because forum does not allow .tsv files.

It is by default sorted on column E, FrequencyMatch, which compares the frequency of letters in the text to known order frequency in English, using my own weighted calculation. Low is better.

The alternative is to use "Similarity" which uses frequency percentage to get similar-but-different ranking. High is better.

Then I also did a quick-and-dirty calc on the top eleven chars ⍽etaoinsrhl which, if char was in correct position in frequency, added integer starting at 11 for space and going down to 1 for ell.

So at the moment there is only 1 text that scores a perfect 66 on the top 11 chars, a piece by John Ward, attached. It's also third on the default FreqMatch sort.

The problem with that is that it only uses 74 of the 95 characters, so using it to optimise for the remaining QXZ`~@#$^&_=+\|]}[{><  is pointless. On the other hand it's only 13k big so provides good fast tests for basic English letters.

The Economic Consequences of Peace (in KLATest) may be a good second choice, it has the joint most characters (only missing `~@#%^\}{><  ) and 59/66 for the top eleven characters. Problem is the large size at 426k which is going to run slowly.

Am open to comments/suggestions about this... both the list and the methodology.

Thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Quick analysis of X8.3
« Reply #1533 on: 2019-Feb-22 10:57 »
Layout X8.3 attached. Has possibly a few unusual key locations.

Did a quick comparison of the English and code tests on KLA, summary of results below. Number of times each layout came first or second. There were occasional wins by others like P_RN, not included here.
CategoryLayout1st2nd
ProseBEAKL 1949
Opted 4 Ergo Alt411
X8.3153
AcademicBEAKL 1511
BEAKL 1913
Opted 4 Ergo Alt41
X8.301
CodeBEAKL 1511
BEAKL 1921
Opted 4 Ergo Alt02
X8.310

I only did the code with code, not Google home page or stripped versions.

Looks like punctuation layer still needs work.

Above also bearing in mind that tests are not necessarily ideal.

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1534 on: 2019-Feb-22 11:41 »
IanDoug, you've been busy ;-)
I'll download your suggested text / corpus, would simplify my life to have ONE good text when testing.

For AltGr, at home I use a Microsoft Sculpt, the Alt key is closer to the middle, so it's not too bad.
I do use a "personal taste" version for this layer, and at work I use a 'wideRight' version, where I move my hand one key to the right to access the Alt key (not perfect, but less tension in the fingers / forearms)

I have gone through the whole "I need to get a 'real' kbd'" thing in the past, and tried to compromise and bought a set of two SmartYao half-keyboards and was disappointed, I don't use them.
I just checked and for me in $CAD it is $500 for an ErgoDox EZ (assuming free shipping applies to Canada hehe).
I am afraid of spending that much for a keyboard without being able to try it out 1st !!
I haven't found a place that have any of these keyboards available here around Montreal.

The new X8.3 sounds interesting, I'll try it out. Seems like a contender for BEAKL19 ;-)

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1535 on: 2019-Feb-22 14:37 »
For AltGr, at home I use a Microsoft Sculpt, the Alt key is closer to the middle, so it's not too bad.
I have gone through the whole "I need to get a 'real' kbd'" thing in the past, and tried to compromise and bought a set of two SmartYao half-keyboards and was disappointed, I don't use them.
I just checked and for me in $CAD it is $500 for an ErgoDox EZ (assuming free shipping applies to Canada hehe).

Ouch... was thinking you may look at Kinesis and memory banks said they were around US200 but I see they're now US350 which is rather steep.
In Rand terms that's around R5000 before postage/VAT ... but on the other hand I had to plonk that down the other day for a hard drive (okay, it WAS an 8TB drive, but still) and it did hurt. At least from my side these things are "business expenses" rather than household expenses. So get a bit of a tax break. But still would turn the dollars over many times before spending it on a keyboard I may not like.

The guys generously offering to buy from Amazon and import on my behalf have prices all over the dial for these. Way overpriced.

Screenshot from ads on Google search result page.


philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1536 on: 2019-Feb-22 16:41 »
Ouch... was thinking you may look at Kinesis and memory banks said they were around US200 but I see they're now US350 which is rather steep.
In Rand terms that's around R5000 before postage/VAT ... but on the other hand I had to plonk that down the other day for a hard drive (okay, it WAS an 8TB drive, but still) and it did hurt. At least from my side these things are "business expenses" rather than household expenses. So get a bit of a tax break. But still would turn the dollars over many times before spending it on a keyboard I may not like.

The guys generously offering to buy from Amazon and import on my behalf have prices all over the dial for these. Way overpriced.

Screenshot from ads on Google search result page.
For a hard drive it seems more justifiable ;-)
There is an ad for an advantage2 on sale @ $500 CAD ($45 off), which means around $575 w. taxes, assuming free shipping.
In this case close to the price of the ErgoDox EZ .. hehe I can still hear my colleague "I Paid $20 for my kbd, works fine"

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1537 on: 2019-Feb-22 17:46 »
hmm, checked Xah Lee's ergo kbds pages again and some of the links I had saved in the past when looking into buying a better kbd.

There is the XBows kbd that is a bit more affordable and could be an interesting alternative to a *full* ergo kbd.
They currently have a special w. free shipping on the "non led" version for $150 US
(nice, I don't want to pay for flashing lights on my kbd .. even though it is pretty cool hehe)

https://x-bows.com/products/copy-of-x-bows-mechanical-ergonomic-keyboard-wrist-rest-combo-free-shipping-worldwide
That should be < $300, at my door, hhmmm :-)

"I consider X-bows keyboard to be one of the top 4 best ergonomic keyboards available." - Xah Lee (I kind of assume he knows his stuff though ;-) )

Seems like good quality, I like the small size too (no room on my kbd / mouse tablet)
I don't particularly like the Alt keys .. nor the Backspace, Enter keys in the middle.
The normal Enter at the right seems far though !?

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1538 on: 2019-Feb-22 19:15 »
oops, just realized this deal is not available with Silent Brown (no Silent anything available !)
price goes up from $150 to $230 US  :-[ :'( ;D

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1539 on: 2019-Feb-22 19:28 »
There is the XBows kbd that is a bit more affordable and could be an interesting alternative to a *full* ergo kbd.
They currently have a special w. free shipping on the "non led" version for $150 US
https://x-bows.com/products/copy-of-x-bows-mechanical-ergonomic-keyboard-wrist-rest-combo-free-shipping-worldwide
"I consider X-bows keyboard to be one of the top 4 best ergonomic keyboards available." - Xah Lee (I kind of assume he knows his stuff though ;-) )

I don't particularly like the Alt keys .. nor the Backspace, Enter keys in the middle.
The normal Enter at the right seems far though !?

If I remember correctly we did discuss the X-Bows a while back but can't find the posts now ... search finds nothing as does Google.

I have a full size picture here: https://www.keyboard-design.com/kb-images/x-bows-scaled.jpg

Xah and I have some differences of opinion on some things.... for X-Bows, I think the odd location of some keys would annoy me. Like Insert.

The layout is some sort of mashup between classic MS Natural and new 10-key-less designs, but with some curious design choices (IMHO).

Would definitely suggest you try before you buy.... must be a dealer somewhere that stocks them.

I see online reviews tend to be positive, but think they may be comparing against ANSI rather than some better form factor.

philippe.quesnel

  • Valued Member
  • ***
  • Posts: 106
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1540 on: 2019-Feb-22 20:20 »
Would definitely suggest you try before you buy.... must be a dealer somewhere that stocks them.
Hehe the voice of reason !
I just can't seem to find a place that has these kind of keyboards

I see online reviews tend to be positive, but think they may be comparing against ANSI rather than some better form factor.
Yes, I am surprised when I read a kbd review and they say " I don’t touch type or use the home row technique", ooops !!!

Thx yet again for your help

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1541 on: 2019-Feb-23 01:50 »
Yes, I am surprised when I read a kbd review and they say " I don’t touch type or use the home row technique", ooops !!!

When there are too many positive reviews and not enough dissenting voices then it's either the greatest thing since sliced bread, or maybe "bought" reviews.

When I see new boards like that my head starts evaluating it on various tracks... starting with how KLA would score the layout, based on distances and angles between the keys. I also look at what they did with the non-30 keys like the navigations, Alt/Ctrl etc. Is it just another unergonomic slab? Was it an attempt to be pretty rather than hand-friendly? etc.

We look at keyboard design differently to others, and people like Den who play games have a whole different set of criteria.

I admit I didn't read the reviews, just noted the "4 stars" and "5 stars" as shown by Google search results, and got to wondering....

I think properly trained touch typists would not like X-Bows because they will have trouble adjusting. Hence you need reviews from non-touch-typists and hope that the bulk of your target market also are non-touch-typists. So board is not aimed at professional market.

I’m sure most people would throw their hands up in horror at having to learn a new keyboard ... getting to grips with ANSI QWERTY was hard enough :-)

Mmmm interesting observation: Firefox spell-checker does not like I'm but is happy with I’m ... you can't see the difference in this font but bad one is ASCII single quote and second one is typographic single quote.
Doesn't like Mmmm either :-)

But can't with ASCII single quote is fine.... odd.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
English prose analysis
« Reply #1542 on: 2019-Feb-23 08:08 »
FWIW, here is analysis of the various English prose-like files I've looked at for use as testing corpus.

On the default sorting (FreqMatch, low is better), you will see Alice pretty low down, as well as various texts I introduced like Jonathan or Declaration of Human Rights etc.

Guide to sorting:

Similarity: high is better
FreqMatch: low is better
TopEleven: high is better
CharCount: high is better

Texts attached, some may be under copyright ... sources include Gutenberg as well as assorted posts around the web.
They have been "cleaned" to be suitable for KLA ... unneeded top and bottom matter removed, non-ASCII chars fixed, converted to UTF-8 with Linux line-endings, etc.

Cheers, Ian



iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Another quick comparison
« Reply #1543 on: 2019-Feb-23 08:14 »
FWIW, similarly to other post above, I ran the 21 tests I'm evaluating as English corpus against other layouts on KLATest, including BEAKL 19.

Some of these duplicated the ones already done, but included them in case Den's posted version differed slightly.

Summary of results:

CategoryLayout1st2nd
ProseBEAKL 1954
Opted 4 Ergo Alt313
X8.3123

Results on Patrick or Den 1 or Den 3 will be different ...
There were others that got occasional first or second, not included, hence columns add to 20 rather than 21.

Cheers, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1544 on: 2019-Feb-23 12:25 »

https://x-bows.com/products/copy-of-x-bows-mechanical-ergonomic-keyboard-wrist-rest-combo-free-shipping-worldwide
That should be < $300, at my door, hhmmm :-)


Massdrop has the Kinesis on special at USD300.

https://www.massdrop.com/buy/kinesis-advantage2-lf

I see that's around 400 CAD ... odd, I thought your currency was stronger than USD.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1545 on: 2019-Feb-23 14:20 »
@Den

Is there somewhere where you have documents the various changes to KLA?

Was poking around trying various CSS files, in particular the one from Materialize:
https://raw.githubusercontent.com/Dogfalo/materialize/v1-dev/sass/_style.scss

I added Seelpy1.22 back to the test panel.

On KLATest with that input, Seelpy wins by a mile (BEAKL 15 comes second at +20% ....) ... but when I run same test on KLA 3 then Opted4 wins and Seelpy is down the list after other recent layouts.

Contrawise, if I run The Economic Consequences of Peace, Seelpy comes stone last on KLAtest ( at +35% behind X8.3) while on KLA 3 it is +40 from X7.1.

Contrawise 2, back in the days of Den 1, this layout was one of the top scorers on just about anything.
https://www.keyboard-design.com/best-layouts-by-average-ranking.html

The changelog on the About page doesn't go into much detail.... :-)

So trying to understand what you're changing between versions .... I have general idea up to Den 3 but no idea on KLAtest.

Thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Other possible input texts
« Reply #1546 on: 2019-Feb-24 15:30 »
Hi

Poked around Gutenberg again.

Some new texts, features as follows:

from-paddington-to-penzance.txt: second best on FreqMatch, 84 chars, first 11 chars okay if you split vowels/consonants by hand.

abc-of-gothic-architecture.txt: 79 chars, first 11 chars okay for vowels if you split vowels/consonants by hand, the r/s balance is badly out.

anatole-france.txt: 78 chars, the first 11 generally okay except r/h swapped

the-hastings-road.txt: 82 chars, first 11 okay if you split vowels/consonants by hand

vaccination-a-delusion.txt: 82 chars, best score on similarity, first 11 chars perfect.

english-hours.txt: 78 chars, first 11 okay if you split vowels/consonants by hand

throne-makers.txt: 80 chars, first 11 nearly perfect (l/d swapped).

Many others painfully cleaned up only to be rejected ....

Cheers, Ian




Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1547 on: 2019-Feb-24 16:22 »
check commit logs for changes:
https://bitbucket.org/Shenafu/keyboard-layout-analyzer/commits/

currently KLAtest over-penalizes the thumbs. although my local version brings it back down.

iandoug

  • Hero Member
  • *****
  • Posts: 941
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1548 on: 2019-Feb-24 16:47 »
check commit logs for changes:
https://bitbucket.org/Shenafu/keyboard-layout-analyzer/commits/

Added link on my Tools page (for my future reference :-)  )
https://www.keyboard-design.com/tools.html

currently KLAtest over-penalizes the thumbs. although my local version brings it back down.

So noted, might explain a few puzzles :-)

Thanks, Ian

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1240
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1549 on: 2019-Feb-24 22:18 »
some new insights after testing new layouts.

feels better to start a bigram from the bottom row and end at top row than vice versa. it's related to how stretching feels better than curling, so that you end a bigram with a stretch rather than a curl.

thus I made changes to your X8-3, which coincidentally also scored slightly better overall:
Code: [Select]
X8-3A

qyou   vdrcz
hieaj  gtnsp
 x k   wmlfb

 

1 Guest, 1 User