Author Topic: Balanced Keyboard Layout  (Read 45956 times)

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1100 on: November 01, 2017, 07:22:41 AM »
http://userinterfaces.aalto.fi/how-we-type/

Those Finns have some other interesting papers. Looks like "keyboard layout" is one of their research areas...

http://userinterfaces.aalto.fi/papers/


Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1101 on: November 01, 2017, 11:45:34 AM »
counting Whole words can be misleading. You can Type 10 keys on the same Hand, then 11th key on the opposite hand, but it doesn't count.

We should be counting any sequences of 4 or more consecutive key presses on the same hand. Ideally also take into account punctuations and numbers, if feasible.

Support me on Patreon

I saw. I conquered. I came.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1102 on: November 01, 2017, 12:52:45 PM »
counting Whole words can be misleading. You can Type 10 keys on the same Hand, then 11th key on the opposite hand, but it doesn't count.

We should be counting any sequences of 4 or more consecutive key presses on the same hand. Ideally also take into account punctuations and numbers, if feasible.

Yeah have been thinking similar thoughts... words are only half the story, need a comma now and then.... :-)

So I think I'm going to put the results of these experiments up on my site (create a section for Alternative Ways of Comparing Layouts or somesuch) and then go back to the drawing board ....

Will ponder your suggestion as well...

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1103 on: November 01, 2017, 01:44:33 PM »
-- new tab: words
-- moved homeKeyWords, homeBlockWords chart to 'words' tab
-- new chart: singleHandWords
-- new score category: words via singleHandWords & homeBlockWords
-- synopsis of text

Got quite a fright when I saw the scores. Thought I was imagining things or my memory was going.
Good work.

Question: on Words tab, you have "Words" and "Total" columns... what's in the totals column?

One of the things I noticed when playing around with the various word metrics was for example on HomeBlock words, from Scrabble list:

X7.1H Ergolinear   youmrcieatnspgdlfh   154000
McGunnigle Peoples rehomed   ypcludsathemgfonir   154000

The second layout is an old layout that was patented, I shifted home keys one column inwards (else LH home would be JBSA instead of BSAT)

It co-incidentally has the same home-block keys as X7.1H so scores the same number of words.

Have attached it.

However the order is different, and it features an awkward row jump for N-D, so typing AND is not pleasant. (AND on this layout is same keys as FNI on QWERTY).

So possibly the next thing to put in the mix is a look at row-jumps. Particularly same-finger row jumps.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1104 on: November 02, 2017, 02:05:49 AM »
Question: on Words tab, you have "Words" and "Total" columns... what's in the totals column?

Total accumulates the length of the words, or total letters counted. thus longer words will more quickly accumulate higher totals.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1105 on: November 02, 2017, 12:36:59 PM »
You busy with something?

Was trying to test Marsan layout. Worked fine on Den 1.


iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1106 on: November 02, 2017, 12:47:18 PM »
mmm. I switched Marsan out for another preset, then it works fine.
Ah. Too many question marks. And a few other things. Fixed, all working okay now.

Layout attached. Tried to load as close as possible to source, could not make out some things, and had to add some punctuation not shown on diagram. I think punctuation could be improved.
Does better than Maltron on KLAtest. Maybe the thumbshift helps.

Possibly C should be on thumb. I would swap it with G but that still leave you with CK on same finger. Maybe on typewriter you would move index to C then hit K with middle.
« Last Edit: November 02, 2017, 12:50:27 PM by iandoug »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Klatest scoring model
« Reply #1107 on: November 03, 2017, 03:26:39 PM »
Hi

@Den

Was tired and not in the mood for work today to spent a lot of the time playing with Essie 3. Still a work in progress, it beats X7.1H on some things (think KLE) but still second on Putin (and most everything else), which I use because letter frequency similar to English General.

However I find myself doing things which I know don't make sense, like swapping z with " because that puts z on "home block" which boosts the "words on home block" score, which improves the overall score.

But from experience I will use " a lot more than z and it should be more convenient.

So I think the contribution from HomeBlock Words needs to be tuned down a bit. Or else you need to include Pinky Home as part of HomeBlock.

I actually think pinky is getting punished too much at the moment. I know it's an issue for you and part of your model, but possibly the metrics need a little tweaking?

Current WIP attached. I'm not really happy with some of the things, at the moment I'm just trying to win on KLatest....

The home block word counts are quite insane.

Cheers, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Oh, those experts
« Reply #1108 on: November 04, 2017, 07:58:41 AM »
Remember children, don't ever learn to drive an automatic car, because then you'll forget how to drive a manual car, and don't learn to drive a manual car because then you'll forget how to ride a bicycle, and don't ever learn to play the piano because the keyboard is completely different to a computer keyboard and you'll forget how to type.

A rant from 2014.

https://arstechnica.com/gadgets/2014/05/storing-two-keyboard-layouts-in-your-brain-at-once-its-possible/

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1109 on: November 04, 2017, 02:05:11 PM »
Re typing styles and use of pinkies, I think I posted the video here, or at least an article about it, but here's the Official Page as well as the conference paper, which I found while trying to find the page I saw the other night which maintained "If you are not using all eight fingers, you are not being efficient" or words to that effect. Google won't show me now.
http://userinterfaces.aalto.fi/how-we-type/

Started reading this, am at about page 4. Some of their findings are "disturbing" in the sense that they break some assumption inherent in KLA and other layout/typing analysis programs.

For example:
1. fastest typists were NOT touch typists.
2. bigrams fastest on alternate hands, not same hand, and even faster than typing same key twice

re (1), KLA assumes perfect touch typist who returns fingers to home when not needed. And always uses 'correct' finger for key.
re (2), there is much talk in the layout optimising community about the desirability of inward rolls... but it would be faster to alternate hands, not do rolls...

re (2), current designs split vowels and most consonants (except sometimes H and D ...) onto different hands, so that makes AND or THE problematic, not to mention IOU.

Coding to ignore/work-around (1) is non-trivial. Although study says best results come from typists always using same finger for same letter, does not specify (so far) an algorithm for which finger on which key, I assume this varies by user, which is impossible for us to deal with fairly.

I hereby retract my assertion that speed=distance/time as far as layout speed is concerned.... :-)

On the other hand the paper mentions some inter-key timings for different situations, in theory it may be possible to work out an Estimated Shortest Time for a given piece of text, and convert that into an estimated WPM....
It would mean that users of AltGr layouts operating at max speed would need to co-ordinate two key presses (thumb+one) within say 140 ms == about 1/7 seconds.

Much to ponder :-)

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1110 on: November 05, 2017, 01:15:22 AM »
KLAtest scoring is teetered around a baseline average. It may seem scores are unfair, but that's only because those layouts are far worse than the average baseline.

Quote
I find myself doing things which I know don't make sense, like swapping z with " because that puts z on "home block" which boosts the "words on home block" score, which improves the overall score.

z is so rare (in English) that it shouldn't improve home block so much more than it negatively affects the other scores, like distance and finger usage.

the secondary issue is your selection on an imperfect corpus which has more z than ". so obviously that will mislead you to wrong solutions.

including pinky in the home block won't solve your particular case, nor affect anything in general. that's only moving the goalpost, moving the baseline, but the scores would stay relatively unchanged.

Consider the case between Arensito and Colemak. Both have AO on the pinkies. But Arensito maintains a high (good) score on Home Block Words category, however Colemak does poorly here. Suppose we include the pinky into home block. We can surmise that Colemak will perform better, but so will Arensito. Relatively speaking, Colemak will still score worse than Arensito.

NO pinky home block

Arensito 28.81
Colemak 35.74

WITH pinky

Arensito 4.48
Colemak 23.16

Relatively speaking, Colemak will actually lose points compared to Arensito. Arensito improved by 24 points, Colemak only 12. But that's not the whole story, since we have to readjust the baseline so that Words category is weighted fairly against the other categories. This means moving the baseline in the opposite direction of the shift in average scores. If the average score decreases, the baseline shifts up. In this case, probably end up with Arensito around 7.x and Colemak around 37.x.
« Last Edit: November 05, 2017, 01:54:59 AM by Den »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1111 on: November 05, 2017, 02:02:56 AM »
KLAtest scoring is teetered around a baseline average. It may seem scores are unfair, but that's only because those layouts are far worse than the average baseline.

You lost me there :-)

z is so rare (in English) that it shouldn't improve home block so much more than it negatively affects the other scores, like distance and finger usage.

Was pondering that exact thing (rarity of z) ... but it's much more common in American English than British English. We tend to use s where Yanks uze z ... (sorry fingers insisted on doing that twize so I left it ...)

Which had me wondering which version the UN uses ... given that it is in USA, but British English is more widely spoken...


the secondary issue is your selection on an imperfect corpus which has more z than ". so obviously that will mislead you to wrong solutions.

Yeah... I think I've pushed Essie 3 as far as it will go on Putin, it's still miles away from X7.1H on Alice, presumably the effect of different language use and much more single quotes, ? and ! messing up things. Current version does well on KLE though.

I'm starting to feel that Klatest makes it impossible for CLP/SC layouts to win at English. Presumably a side-effect of the algos used rather than by design.
But it may mean that CLP/SC as concept was bad idea.


iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Proposal for layout file names
« Reply #1112 on: November 06, 2017, 02:16:46 PM »
Hi

Want to propose a standard format for layout file names.

Patrick originally used format.layoutname where format was one of standard, european, ergodox, and layoutname was a text string which may have been meaningful or not, but not in any standardised format.

Den has switched that around a bit to layoutname.format.json, where format is as above, including ergolinear and matrix, and names are basically as per Patrick.

So we end up with things like
colemak.standard.json
Colemak-Mod-DH-full.standard.json
colemak-estonian.european.json
colemakjjt2.ergodox.json

etc. Inconsistent capitalisation is another issue.

So I was thinking something along the lines of

Family_Variant_ModdedBy.language.format.json

where format would be one of ansi, iso, ergodox, matrix, ergolinear
eg
Colemak_DH_mod-Joe-Bloggs.en.ansi.json

There may be more than one Modder. Am using underscores where parts may be variable in number and dots where they are fixed. To aid parsing.

Dvorak.en.ansi.json
Dvorak_Caps-Backspace.en.ansi.json
Dvorak_mod-Kaeteker.en.ansi.json
Dvorak_Programmer.en.ansi.json
Dvorak_Programmer-Ordered-Numbers.en.ansi.json
Dvorak.fr.iso.json
BEAKL_8_mod-Ian.en.matrix.json

Comments? Unnecessary?

Then need to worry about getting sane names in the files as well... they're currently a mishmash of all sorts of things.
Might be simplest to use exactly the same filename as the layout name.

Thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
WordCount metrics
« Reply #1113 on: November 07, 2017, 05:09:13 AM »
Hi

Okay finished doing the sums on the various word count metrics. Am in the process of creating pages for them on my site.
Did HomeKey words, HomeBlock words, EasyBlock words, and OneHand words, all based off the Scrabble word list, with minimum word length of 2 (ie the entire list).

I then "normalised" the scores relative to the best performer, who got 100%, everything else got the relevant percentage of that.

If we average the HomeBlock and OneHand scores (ie one positive scoring, one negative scoring), we get the Top Ten:

LayoutAverage
Modified BEAKL54.75
Essie 3 Ergolinear wip52.6
Marsan English51.51
BEAKL 5 Matrix44.68
BEAKL CLP 042.23
BEAKL Opted36 (+Ergo)41.67
Einbinder Orthogonal   41.67
US Pat 3,929,216 mod Ian40.44
X7.1H Ergolinear40.13

Which is curious because apart from BEAKL variants and from me, the other three are from old layouts, some patented decades ago.

Maybe their "typewriter" design mindset co-incidentally resulted in layouts similar to what Den is now aiming at, with little pinky use.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Guptal patent
« Reply #1114 on: November 07, 2017, 12:07:14 PM »
Patent from 2007.

Attached. Will try to figure out how to load this on Matrix or Ergodox.

There is an ISO version too, but the diagrams in the patents are rather crappy and difficult to make out. Else I haven't found the right version yet.

WIPO: https://patentscope.wipo.int/search/en/detail.jsf?docId=WO2007013095&recNum=1&tab=Drawings&maxRec=&office=&prevFilter=&sortOption=&queryString=

Espace: https://worldwide.espacenet.com/publicationDetails/claims?CC=WO&NR=2007013095A3&KC=A3&FT=D&ND=&date=20070726&DB=&locale=#  (the attached was from a PDF I downloaded here)

(There is a text description of the letters on each row on ISO as well. )
« Last Edit: November 07, 2017, 01:54:15 PM by iandoug »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Finger tip pain
« Reply #1115 on: November 07, 2017, 12:26:05 PM »
@Den

One of my problems with my fingers is that the top joint of my right middle finger is very sore. It finally dawned on me why ... because I use it so much for copy-paste operations (with ctrl or shift on right hand) but the ANGLE is all wrong.

When I type on the alphas, my fingers are curved and press straight down. Like they should.

But when I hit ctrl-insert to Copy, thumb is on ctrl, and tip of middle finger is extended horizontally, and then from that angle presses down on Insert.

Which very effectively forces/torques the tip of the finger up, twisting it in a direction it was never designed for. No wonder it's sore after years of such abuse.

And then of course (serendipity and all that) tonight I stumble across a paper on whether keys should be set at an angle for precisely that reason ... to reduce finger pain.

Which got me thinking about Maltron/Kinesis and their odd key angles.

So I stare at hi-res pics of Kinesis.

Then I realise why Den can't do the mouse on the left hand ... because crtl-c/x/v are on left, but so is Insert key. That's rather short-sighted design from my point of view...

So in any physical design I come up with, I must either make sure keys are easily reachable and pushable straight down, or angle them somehow ....

Copy of said paper attached. Also another one which has some interesting pictures etc.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Finger tip pain
« Reply #1116 on: November 07, 2017, 08:12:59 PM »
Copy and paste are extremely common, repetitive functions across every application. For best effect:

- They should be activated by a single finger press; not combination.
- They should be placed in very convenient locations.

Single finger press means your whole hand is free to access the key, not bound to painful positions in awkward  combinations.

I put them on the mouse button activated by the thumb. My right hand is almost always on the mouse anyway. Using the mouse to select text, followed by mouse thumb keys to copy and paste, allows nice sequence.

If I intend to paste a lot in a short time and want to do it fast, occasionally I use the left hand to press the Escape key that has been replaced by paste function. Again no combination.

(It's not really Escape any more, whose function I have moved to the left shift location. The far corner key where Escape originally sat is reprogrammed to useless key--NumpadEnd--and became a macro in AutoHotkey to activate Ctrl-V. thus one-hit the old Escape key would paste stuff.)

Combinations, especially awkwardly caused by keys far apart and hit by the same hand, obviously not good to stretch your hand like that. More reason that more keys on a board are not necessarily better. They become convenience traps: look very convenient, but not necessarily ergonomic. Another reason to the distance theory has merit even outside of touch typing.

« Last Edit: November 08, 2017, 04:37:20 AM by Den »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Finger tip pain
« Reply #1117 on: November 07, 2017, 10:17:16 PM »
Copy and paste are extremely common, repetitive functions across every application. For best effect:

- They should be activated by a single finger press; not combination.
- They should be placed in very convenient locations.

Single finger press means your whole hand is free to access the key, not bound to painful positions in awkward  combinations.

Mmm so I'm thinking of remapping right Windoze and Menu buttons to copy and paste. "Cut" is not so bad where it is now, and not used so often. Suppose could make shift-Windows do Cut if necessary.

Have just realised now that I will still have the same problem with the Ctrl-Home combo, which I use to go to the top of the page in the the browser, editor, etc....

I use trackball with 4 buttons, but three are for mouse (left, middle, right) which only leaves one free for copy/paste ops, so will be better if I can get the keys on keyboard remapped.
Really must get going with my own keyboard... :-)

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: WordCount metrics
« Reply #1118 on: November 08, 2017, 04:25:36 AM »
If we average the HomeBlock and OneHand scores (ie one positive scoring, one negative scoring), we get the Top Ten:

LayoutAverage
Modified BEAKL54.75
Essie 3 Ergolinear wip52.6
Marsan English51.51
BEAKL 5 Matrix44.68
BEAKL CLP 042.23
BEAKL Opted36 (+Ergo)41.67
Einbinder Orthogonal   41.67
US Pat 3,929,216 mod Ian40.44
X7.1H Ergolinear40.13

Which is curious because apart from BEAKL variants and from me, the other three are from old layouts, some patented decades ago.

Maybe their "typewriter" design mindset co-incidentally resulted in layouts similar to what Den is now aiming at, with little pinky use.

Well, Einbinder does looks similar to 3-thumb-letter layouts like U_RN.

Marsan is really outlier. It doesn't have great home-block score, but it does have really great one-hand score; guess that evens out.

re: HomeKeys and HomeBlock

If we count every letter typed by keys in those regions, we get the percentage of letters typed by those keys. Perhaps a more practical metric--easier to measure and compare. For example, typing short words like 'a', 'or' on some layouts requires pinkies. Unfortunately this is not penalized if we only account for lengthier words.

OneHand stays to measure strings of three or more letter consecutively typed on the same hand. That is to say that we accept it as normal, acceptable to type single or pairs of letters on the same hand.
« Last Edit: November 08, 2017, 04:34:22 AM by Den »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: WordCount metrics
« Reply #1119 on: November 08, 2017, 06:56:55 AM »
Well, Einbinder does looks similar to 3-thumb-letter layouts like U_RN.
Marsan is really outlier. It doesn't have great home-block score, but it does have really great one-hand score; guess that evens out.

re: HomeKeys and HomeBlock
If we count every letter typed by keys in those regions, we get the percentage of letters typed by those keys. Perhaps a more practical metric--easier to measure and compare. For example, typing short words like 'a', 'or' on some layouts requires pinkies. Unfortunately this is not penalized if we only account for lengthier words.
OneHand stays to measure strings of three or more letter consecutively typed on the same hand. That is to say that we accept it as normal, acceptable to type single or pairs of letters on the same hand.

Some time after I posted those averages it dawned on me that they're actually rather low. Considering that a layout that was very good at both metrics would have gotten say 80+, the best we have is 54 which is decidedly average.
Means it was half as bad/good as winner in each metric, or very good at one and very bad at the other.

The first few at HomeBlock words were like this:

Essie 3 Ergolinear wip      267751
BEAKL CLP 0      182070
Essie 2      158743
X7.1H Ergolinear      158321
...
Einbinder Orthogonal   13886
...
Marsan English   8076

and for OneHand

Marsan English 93
Modified BEAKL   109
Einbinder Orthogonal   119
US Pat 3,929,216 mod Ian   119
BEAKL Opted36 Ergo   141
BEAKL 5 Matrix    141
...
BEAKL CLP0 565
...
QWERTY    5291

So Marsan starts with 100% on one test, which helps a lot... :-)

I think the fact that the SC layouts get such way-above-others numbers on the HomeBlock words are skewing the calculations a bit. Maybe should do square root of those numbers rather than use as is.
Will try that and see what surfaces.

I agree with your 3-letter cutoff in KLA for one-handed words, I just used 2-letter words in this exercise in case someone complained I wasn't comparing apples to apples.
I can run the whole thing again with say 4-letter words but results should be similar. Think using square of HomeBlock words will change things slightly because the normal layouts won't get such a lower percentage compared to SC layouts.

Still find it fascinating that such different methodologies produces an order largely in line with finger-based metrics. I suppose it's partly because HomeBlock by definition excludes pinkies. The Colemak fanbois won't like that.

Speaking of Colemak, the other day was on about page 30-something of results from Google, which sent me to the Colemak forums, where you were having a big debate with Michael Dickens... he was in the early stages of MTGap. Had no idea you two were so close :-).. Or that you've been doing this for so long :-)

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: WordCount metrics
« Reply #1120 on: November 08, 2017, 07:34:27 AM »
I think the fact that the SC layouts get such way-above-others numbers on the HomeBlock words are skewing the calculations a bit. Maybe should do square root of those numbers rather than use as is.
Will try that and see what surfaces.

Ended up taking square roots for both metrics.

Modified BEAKL   70.77
BEAKL 5 Matrix   64.79
Essie 3 Ergolinear   63.15
BEAKL CLP 0   61.52
BEAKL Opted36 Ergo   61.46
X7.1H Ergolinear   61.44
BEAKL 10 Matrix   59.29
Marsan English   58.68
BEAKL 8 Matrix   58.06
BEAKRAK   57.32
BEAKL 7 Matrix   57.09
BEAKRAK mod Ian   56.41
BEAKL 8 Matrix mod Ian   56.1
BEAKL 9 Matrix   55.69
Einbinder Orthogonal   55.59
BEAKL 7 Matrix Mod Ian   55.51


There's something about BEAKL.... :-)

Modified BEAKL is a variant I found on the web, by ZeptBytes.
https://github.com/ZeptByteS/keyboard_layout_analyzer
Has another page somewhere as well (in Chinese or Korean IIRC)
« Last Edit: November 08, 2017, 12:19:49 PM by iandoug »

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1121 on: November 08, 2017, 10:46:45 AM »
Colemak forums used to be popular to discuss alternative layouts. But it stagnated since AdNW. which the Germans / Europeans now much more receptive to discovering better layouts. Even France considering changing national layout to Bepo, a Dvorak like layout. Too bad I don't understand their languages to join in their discussions.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1122 on: November 08, 2017, 12:01:48 PM »
Even France considering changing national layout to Bepo, a Dvorak like layout.

Yeah, have seen stories about that. I hope they do, it will be the first major onslaught against QWERTY and cousins in the west.
Hopefully they look at the ISO form factor too at some point. Else it's just deck chairs on the Titanic.

If they really wanted to improve productivity they could start by reforming their spelling... what I hated most about my 6 months of French in high school.
Simpler spelling == less typing. Shorter books. Etc.

Actually saw a proposal the other day for a 20-year plan to reform ANSI/ISO ... something along the lines of moving the middle row 1/4 key to the right (and keep offset of top row at 1/4 of middle). Then give people a few years to adjust. Then make another move, which will result in two rows becoming ortho. Wait a few more years....

And ignore the ANSI/ISO standards bodies having a cadenza....

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Guilty as charged
« Reply #1123 on: November 09, 2017, 12:02:46 PM »
Will the accused please rise?


from http://www.rutherfordjournal.org/article010105.html

Skip to the "Keyboard Design: Universal, Ideal, and Simplified" part.

Quote
According to Beeching, in 1905 a “large international meeting” was held on the issue of keyboard standardisation:
Quote
    At that time various keyboards—certainly more efficient than the one devised by Sholes and used today—were put forward as alternatives. The battle raged backwards and forwards. Nobody could agree on what a new keyboard should be, but the biggest opposition came from the teachers of typing as it still does today. They wanted things to remain as they were, and they are still most reluctant to change their methods and learn all over again.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Font on Klatest
« Reply #1124 on: November 09, 2017, 12:50:18 PM »
@Den: possible to swap Lato out for Noto?

I think when Patrick built KLA, Lato was best Google had, but now there is Noto... which features better differentiation between cap I and lower l. Which helps when dealing with SC layouts.

Thanks, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: WordCount metrics
« Reply #1125 on: November 10, 2017, 06:32:31 AM »
Ended up taking square roots for both metrics.

Finally dawned on me the problem with the one-hand metrics, as regards AltGr layouts ... I was ignoring which hand the needed thumb was on.
So I need to see where the Shifts and AltGrs are, how they are used, and then only use letters accessed by them if on same hand.

Biggest effect is going to be on the SC layouts.

Dunno if Den is using same methodology, or doing what I was doing previously and just looking which hand is used to press the key ...

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1126 on: November 10, 2017, 09:10:33 PM »
BEAKL 9 (Personal speed typing progress)
Website, https://www.speedtypingonline.com/typing-test

Corpus Sight Words (Easy-to-type)
Author bogglesworldesl.com
Type Random Words
Test Results
Speed 76 WPM
Accuracy 96.3 %
Correct Entries 367
Incorrect Entries 14
Fixed Mistakes 14
Total Entries 381
Error Rate 0
Raw Speed 76 WPM
Key Speed 381 KPM
Complete Words 82
Total Time 1 min(s)


Corpus The Fox and the Goat
Author Aesop
Type Fables (Easy to type)
Test Results
Speed 74 WPM
Accuracy 95.7 %
Correct Entries 356
Incorrect Entries 16
Fixed Mistakes 16
Total Entries 372
Error Rate 0
Raw Speed 74 WPM
Key Speed 372 KPM
Complete Words 75
Total Time 1 min(s)


Only been with BEAKL 9 for around 40 days, and already up to 75 WPM (started at around 55-60). I think it's pretty good progress, and shows that this layout is easy and has low learning curve. I don't practice and hone speed typing, but naturally improving by using it for everything. Still make too many errors, though.

I seem to type faster when leaning back, arms stretched out. Elbows at right angles can feel too stiff and not as mobile.
« Last Edit: November 11, 2017, 03:59:59 PM by Den »

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Guilty as charged
« Reply #1127 on: November 10, 2017, 09:30:00 PM »
Will the accused please rise?


from http://www.rutherfordjournal.org/article010105.html

Skip to the "Keyboard Design: Universal, Ideal, and Simplified" part.

Ya of course there's parties interested to maintain QWERTY as the standard.

It's also amazing that people were able to hit 140 WPM with those ancient typewriters with steep rows.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Guilty as charged
« Reply #1128 on: November 11, 2017, 01:10:23 AM »
It's also amazing that people were able to hit 140 WPM with those ancient typewriters with steep rows.

I also find that hard to believe. Maybe they counted actual words back then, and lots were less than 5 letters long....

"So if I had to go and do it"
« Last Edit: November 11, 2017, 01:21:31 AM by iandoug »

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: WordCount metrics
« Reply #1129 on: November 11, 2017, 01:55:26 AM »
Finally dawned on me the problem with the one-hand metrics, as regards AltGr layouts ... I was ignoring which hand the needed thumb was on.
So I need to see where the Shifts and AltGrs are, how they are used, and then only use letters accessed by them if on same hand.

Biggest effect is going to be on the SC layouts.

Dunno if Den is using same methodology, or doing what I was doing previously and just looking which hand is used to press the key ...


For SingleHand words, Shifts and AltGrs are transparent. That is, they don't add or subtract to the handed-ness. They are just helpers to access letters on different layers--even if they are opposite hand. Ultimately the letters are still typed on the same hand, including stuff like rolls, row jumps, same fingers, etc. are still happening with the non-thumb fingers. Therefore effort is expended likewise as if on the same layers.

Seelpy 1.5 still has good Words score on KLAtest despite that. The excellent HomeBlock score compensates for bad SingleHand score, but overall is quite competitive in this category.

re: overall Words scores

Some layouts really stumble on this category. Including Arensito, Colemak, BEPO, etc. As expected from low hand-alternation layouts, which leads to lengthy one-handed strings, up to 10, 12, 14 letters consecutively on the same hand (according to your chart)--that's ridiculously inefficient.


iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: WordCount metrics
« Reply #1130 on: November 11, 2017, 03:14:18 AM »
Seelpy 1.5 still has good Words score on KLAtest despite that. The excellent HomeBlock score compensates for bad SingleHand score, but overall is quite competitive in this category.

Some layouts really stumble on this category. Including Arensito, Colemak, BEPO, etc. As expected from low hand-alternation layouts, which leads to lengthy one-handed strings, up to 10, 12, 14 letters consecutively on the same hand (according to your chart)--that's ridiculously inefficient.

I've been playing with different methods to get an average of HomeBlock+SingleHand.

The more intensely I compress the range (eg taking square, or log base 2) of raw word count, before turning into relative percentage of "best", the better Marsan does. But if we look at other metrics for Marsan, like hand balance, then it does not really compete. So I thought it may be because my "SingleHand" definition was too broad (allowing help from thumb on other hand) which resulted in SC layouts getting large SingleHand word counts and thus poor relative score.

I've tried with various word lengths as well, using 2 or 4 or 5.

Here's result with 5 letters, log base2 on scores before relative percentage of average of metric.
[ to be more clear:
1. count words (eg homeblock or onehand)
2. convert count -> log base 2
3. find average of all from (2) for homeblock, and singlehand
4. find relative percentage for each from (2) compared to the average
5. take simple average (homeblock+singlehand)/2 of each layout's score from (4)
]

Am posting a lot here just to show how far down X7.1 is, and by other finger-based metrics it should be at least in top 10.

Layout Averagehomeblock log 2Relative %onehand log 2Relative %
Marsan English214.678.847503625923691.3954751399431.7917594692281337.93481905291
Modified BEAKL178.811.029455571214113.935226826292.484906649788243.67028518821
BEAKL 5 Matrix168.6211.002033172315113.651951081762.7080502011022223.5918343661
BEAKL Opted36 Ergo167.0610.699304130942110.524733988132.7080502011022223.5918343661
BEAKL Opted36167.0610.699304130942110.524733988132.7080502011022223.5918343661
Einbinder Orthogonal151.449.414097209490297.2484357004312.9444389791664205.64118200586
US Pat 3,929,216 mod Ian147.758.697846691109589.8495061023892.9444389791664205.64118200586
BEAKL 10 Matrix144.8210.968629231143113.306885489923.4339872044851176.32503441745
schizoKBD-shifted-AltGrSpc139.710.559815134958109.083800635193.5553480614894170.30622643631
schizoKBD-AltGrSpc139.710.559815134958109.083800635193.5553480614894170.30622643631
Colibius August mod Ian137.379.527629789180698.4212370355183.4339872044851176.32503441745
HIEAQMTSRN137.210.077692928728104.103436684313.5553480614894170.30622643631
HIEAQDTSRN136.39.9019864264099102.288373368883.5553480614894170.30622643631
schizoKBD-shifted136.2110.586937870117109.363980827463.7135720667043163.04999637648
BEAKL 8 Matrix135.2911.286376120521116.589238248033.9318256327243153.99917712027
BEAKL 7 Matrix134.9111.211563582462115.816418280793.9318256327243153.99917712027
HIEAMTSRN134.8310.10508054982104.386353171093.6635616461296165.27575362615
BEAKL 8 Matrix mod Ian134.4911.131357531224114.987882857483.9318256327243153.99917712027
BEAKL 9 Matrix134.3711.106534928786114.73146323453.9318256327243153.99917712027
X1 Ergo134.310.218626705596105.559294748313.7135720667043163.04999637648
X1 Ergolinear tweaked134.310.218626705596105.559294748313.7135720667043163.04999637648
X1 Ergolinear134.310.218626705596105.559294748313.7135720667043163.04999637648
alice 74.99 WIP134.310.218626705596105.559294748313.7135720667043163.04999637648
a74.29 kle70.07 WIP134.310.218626705596105.559294748313.7135720667043163.04999637648
a74.89 kle68.82 WIP134.310.218626705596105.559294748313.7135720667043163.04999637648
Ian R1 p134.310.218626705596105.559294748313.7135720667043163.04999637648
Ian M2134.310.218626705596105.559294748313.7135720667043163.04999637648
Ian M1 2134.310.218626705596105.559294748313.7135720667043163.04999637648
BEAKL 4 Mod Ian AltGr 3134.310.218626705596105.559294748313.7135720667043163.04999637648
Ian X1134.310.218626705596105.559294748313.7135720667043163.04999637648
BEAKL 7 Matrix Mod Ian134.2611.085950673374114.518826101673.9318256327243153.99917712027
newden mod Ian134.2611.085950673374114.518826101673.9318256327243153.99917712027
BEAKL EZ Matrix133.9611.027393898906113.913929573633.9318256327243153.99917712027
BEAKL Opted4 Ergo Alt13310.841852838729111.997274427733.9318256327243153.99917712027
X1 Ergolinear OU132.699.906134047477102.331218652043.7135720667043163.04999637648
Ian M2 tweak132.699.906134047477102.331218652043.7135720667043163.04999637648
BEAKL Opted1131.9710.642874027514109.941806155263.9318256327243153.99917712027
BEAKRAK 1.2131.119.600827196903699.1773725671563.7135720667043163.04999637648
BEAKL Opted4 Ergo130.9510.446102498792107.909139301193.9318256327243153.99917712027
BEAKL 4130.9510.446102498792107.909139301193.9318256327243153.99917712027
BEAKL Opted4 Ergo Mod Ian 1129.7810.218626705596105.559294748313.9318256327243153.99917712027
BEAKL Opted3129.5710.938041204904112.990908542084.1431347263915146.14487628484
BEAKRAK129.2111.025995388255113.899482838044.1896547420264144.52215022545
BEAKRAK mod Ian128.7910.943976255854113.052218130014.1896547420264144.52215022545
Ian S3128.779.147294070258594.4923362728533.7135720667043163.04999637648
Aus der Neo-Welt128.629.413526083949497.2425359243623.7841896339183160.0072857324
BEAKL Stretch128.119.8955561822098102.221948391993.9318256327243153.99917712027
X6 Ergolinear127.988.99504097468592.9195486708573.7135720667043163.04999637648
Gupta WHRL127.8810.819118066514111.762422270954.204692619391144.00527382847
Ian S2127.158.834191318202191.2579578536143.7135720667043163.04999637648
Ian X4127.158.834191318202191.2579578536143.7135720667043163.04999637648
Ian M3127.158.834191318202191.2579578536143.7135720667043163.04999637648
X5 Ergolinear127.158.834191318202191.2579578536143.7135720667043163.04999637648
X4 Ergolinear127.158.834191318202191.2579578536143.7135720667043163.04999637648
Ian R2 p127.158.834191318202191.2579578536143.7135720667043163.04999637648
OneProduct127.069.6930134043226100.129664036553.9318256327243153.99917712027
HIEAMTSRN-wide126.3510.204517642507105.413546910364.1108738641733147.2917759158
HEIADTSRN126.210.174659072861105.105105312014.1108738641733147.2917759158
HEIADTSRN-wfgv126.0810.152376820269104.874927721924.1108738641733147.2917759158
Balance Twelve125.8410.10508054982104.386353171094.1108738641733147.2917759158
HAEIK125.419.654192512489199.7286408774214.0073331852325151.09747156819
P_RN124.7910.584081267024109.334471871274.3174881135363140.24309878736
P_NR mod Ian124.7910.584081267024109.334471871274.3174881135363140.24309878736
Ian M1122.829.63154758703399.4947168453924.1431347263915146.14487628484
Ian S1122.829.63154758703399.4947168453924.1431347263915146.14487628484
Minimotion122.549.031572493381593.296922408133.9889840465643151.79251281826
X7.1H Ergolinear122.3611.944707881396123.389862088254.9904325867787121.33174859913

So Marsan is below average on HomeBlock, but far above 2nd with OneHand, and ends up on top.

for comparison, the bottom end looks like this, with the worrying inclusion of Amuseum (and Plum Ergolinear, for that matter):


Plum Ergolinear   82.39
McGunnigle Scientific   82.32
Amuseum   82.17
QWERTY row-swap mod Ian   82
Krush 200 simplified   81.79
AZERTY   81.36
QWERTY rehomed   81.07
QWERTY rehomed Ian   80.62
Parkinson   80.1
Savard Vertical Alpha   79.48
Reverse QWERTY   77.59
ABCDEF   74.98
QWERTY typewriter   71.55
TNWMLC (Worst CarpalX layout)   68.42



Plain QWERTY got 83.41.
The various SC layouts are buried in the middle.








« Last Edit: November 11, 2017, 03:30:53 AM by iandoug »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design


iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Internet Letter Layout DB
« Reply #1133 on: November 11, 2017, 02:36:57 PM »
I've put up an initial version of the magnanimously-titled Internet Letter Layout DB at

http://www.keyboard-design.com/internet-letter-layout-db.html

Stuck a link to it on Deskthority layout page as well, dunno if it will last there.
https://deskthority.net/wiki/Keyboard_layouts

Still a lot to fix but I needed to get the task out of the way. :-)
Fixing everything will take time, so data-per-layout will gradually improve and be corrected.
Will also try to switch to using kle-render for the images, but need to get backgrounds working with that tool.



mstacker

  • Member
  • **
  • Posts: 34
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1134 on: November 11, 2017, 02:42:50 PM »
BEAKL 9 (Personal speed typing progress)
Website, https://www.speedtypingonline.com/typing-test

Only been with BEAKL 9 for around 40 days, and already up to 75 WPM (started at around 55-60). I think it's pretty good progress, and shows that this layout is easy and has low learning curve. I don't practice and hone speed typing, but naturally improving by using it for everything. Still make too many errors, though.

I seem to type faster when leaning back, arms stretched out. Elbows too straight can feel too stiff and not as mobile.

Impressive. I'm probably around 30 wpm still. I think it's week 6 on BEAKL 9. Of course, I've only recently switched full-time. Ended up having to switch entirely as my typing on a dvorak kinesis advantage at work has seriously degraded. Due to the 'key confusion,' I'm about 30-ish on both layouts at the moment. Guess, I was unable to maintain both. Incidentally, that is what I wanted to know.

I also find that hard to believe. Maybe they counted actual words back then, and lots were less than 5 letters long....

"So if I had to go and do it"

About the 140 wpm thingy. Hard to imagine. Agree, we would have to better understand the metrics and protocols used.


Matt

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: WordCount metrics
« Reply #1135 on: November 11, 2017, 03:55:51 PM »
for comparison, the bottom end looks like this, with the worrying inclusion of Amuseum (and Plum Ergolinear, for that matter):

Don't be too concerned for Amuseum layout. It was designed for rolls. So in fact it doesn't mind the long one-hand sequences.

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1136 on: November 12, 2017, 12:08:01 PM »
Impressive. I'm probably around 30 wpm still. I think it's week 6 on BEAKL 9. Of course, I've only recently switched full-time. Ended up having to switch entirely as my typing on a dvorak kinesis advantage at work has seriously degraded. Due to the 'key confusion,' I'm about 30-ish on both layouts at the moment. Guess, I was unable to maintain both. Incidentally, that is what I wanted to know.

It helped that I had been using BEAKL EZ prior for a while. The 9 most common letters are in the same places between BEAKL EZ and BEAKL 9. So the transition is easier this time around.

After using Dvorak for years, it began to fit like a glove, second nature. So it'll take yet longer than a month or two to regain the same feeling of mastery with a new layout.

mstacker

  • Member
  • **
  • Posts: 34
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1137 on: November 13, 2017, 07:25:06 AM »
It helped that I had been using BEAKL EZ prior for a while. The 9 most common letters are in the same places between BEAKL EZ and BEAKL 9. So the transition is easier this time around.

After using Dvorak for years, it began to fit like a glove, second nature. So it'll take yet longer than a month or two to regain the same feeling of mastery with a new layout.

Good to know. I figure another 1.5 months. Must say the BEAKL 9 layout does seem pleasing to the hands after a long day of typing.

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1138 on: November 13, 2017, 07:45:33 AM »
Good to know. I figure another 1.5 months. Must say the BEAKL 9 layout does seem pleasing to the hands after a long day of typing.

http://esr.ibiblio.org/?p=4975


mstacker

  • Member
  • **
  • Posts: 34
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1139 on: November 13, 2017, 04:10:31 PM »
http://esr.ibiblio.org/?p=4975

Sounds like this guy really values his keyboard. I guess I can relate.

Matt

mstacker

  • Member
  • **
  • Posts: 34
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1140 on: November 13, 2017, 04:12:21 PM »
Good to know. I figure another 1.5 months. Must say the BEAKL 9 layout does seem pleasing to the hands after a long day of typing.

@Den.

What keyboard(s) do you use to type BEAKL 9 on?

Thanks,

Matt

Den

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1080
  • Selina is my Superstar
    • View Profile
    • Amuseum
Re: Balanced Keyboard Layout
« Reply #1141 on: November 13, 2017, 06:06:47 PM »
@Den.

What keyboard(s) do you use to type BEAKL 9 on?

Thanks,

Matt

Kinesis Advantage2 (cheaper elsewhere)

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1142 on: November 13, 2017, 09:14:27 PM »
Sounds like this guy really values his keyboard. I guess I can relate.

"This guy" is an "internet god". As is Richard Stallman. :-)

https://en.wikipedia.org/wiki/Eric_S._Raymond.
https://en.wikipedia.org/wiki/Richard_Stallman
http://www.catb.org/esr/who-is-ESR.html

I posted it because they're both been typing for decades. :-)

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
fingers etc
« Reply #1143 on: November 14, 2017, 04:11:27 AM »
Interesting idea.

Was browsing papers again. Actually looking for "academic" texts for evaluating keyboards against.

Anyway Human engineering the one-handed keyboard by Z. Eilam (Applied Ergonomics 1989, 20.3,225 -229) has interesting actual measurements for key stagger and thumb offset.

And also suggests that Pinky keys should be about 5mm HIGHER off keyboard than other keys. First time I've seen that idea. It's so that they don't need to move so far to actuate.

Ob screenshot attached.

Also that test subjects preferred that arms to be below plane of keyboard rather than above...

mstacker

  • Member
  • **
  • Posts: 34
    • View Profile
Re: Balanced Keyboard Layout
« Reply #1144 on: November 14, 2017, 11:53:08 AM »
"This guy" is an "internet god". As is Richard Stallman. :-)

https://en.wikipedia.org/wiki/Eric_S._Raymond.
https://en.wikipedia.org/wiki/Richard_Stallman
http://www.catb.org/esr/who-is-ESR.html

I posted it because they're both been typing for decades. :-)
Sorry. A bit short lately. Too much craziness on television and public media. His tone was a bit on the hysterical side so he qualified as "that guy" briefly. Course the article was published years ago. Constructive feedback if you are listening Eric. I've got a couple of Raymond's books on unix system and network stack programming. Really good. Think I saw a photo of him trying out someone's Keyboardio at some point. As I said, I guess I can relate to his sentiment; keyboards are not an accessory. RMS is a little more colorful. I always kinda looked up to Bill Joy. McKusick is kinda cool too, but I digress.

Would love to hear any comments, thoughts, ideas on keyboard layouts, keyboards, etc from all of the above. Thoughts on BEAKL, PR_N, Seelpy or Xn.m would be really cool.

Matt
« Last Edit: November 14, 2017, 01:07:24 PM by mstacker »

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1145 on: Yesterday at 01:58:04 PM »
"This guy" is an "internet god". As is Richard Stallman. :-)

I probably could have worded that better. Problem with the net is you never know how much experience the person behind the handle has, and of course written text lacks the subtle nuances of face to face speech...

Cheers, Ian

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Let the fingers do the walking
« Reply #1146 on: Today at 12:17:45 AM »
Hi

Attached screengrabs from two different papers.

Comments:

1. Conventional wisdom is that middle finger is strongest and gets best letter allocation. However attached shows that index is both stronger at pushing and faster at tapping. So have we been doing it all wrong? Discuss. :-)

2. Smokers tap faster. Presume because they're "wired" due to nicotine stimulant factor? How about that other programmer staple, caffeine?

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
So long, and thanks for all the rolls...
« Reply #1147 on: Today at 01:36:04 AM »
Screengrab from 1964 paper by Fox and Stansfield, they timed "average" typists on conventional typewriter doing common and not-so-common bigrams, on same hand and alternate hands.

Results clearly show alternate hands are faster.

Which again raises the question of the importance of "inward rolls" etc on same hand ... they may feel nice, but are they faster or more efficient that alternating hands?


iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Pinky crowned champ
« Reply #1148 on: Today at 03:46:31 AM »
1990 paper by Lachnit and Pieper, Speed and accuracy effects of fingers and dexterity in 5-choice reaction tasks

They took three groups of people (controls, typists, pianists) and gave them some tests using five coloured buttons on a custom keyboard, and measured reaction time and accuracy.

Screengrabs of reaction time (RT) and accuracy attached. I stuck my screen ruler on the RT chart just above Typist Pinky bar, for easy comparisons.

Crux of their findings:

Quote
There were clear speed differences between the fingers of the preferred hand; thumb
and little finger showed significantly shorter RTs than those of the other three fingers,
while there was no difference within the two groups of fingers. These effects were
independent of dexterity. Pianists were faster than controls and typists, but there was
no difference between controls and typists although the latter might have been expected
to have greater skills. These results were not confounded by differences in speed accuracy
trade-off between groups. The independence of finger differences and
dexterity suggest that the faster RT of the pianists must be due to an overall
enhancement.
Analyses of accuracy showed a clear superiority of the little finger in all groups, it
was also the most reliable and showed the lowest rates of false alarms.

So while the pinky may be "weaker" it clearly has other skills (reaction time and accuracy) to compensate. Not sure how this will impact current BEAKL theory... :-)

iandoug

  • Hero Member
  • *****
  • Posts: 704
    • View Profile
    • Keyboard Design
Re: Balanced Keyboard Layout
« Reply #1149 on: Today at 06:11:17 AM »
That's where some of the arbitrary penalties can come in. On a flat keyboard, going below what is typically the home row usually requires either curling and extending middle or ring fingers, or hand movement, while the row above home just requires extending (the normal number row is going to vary a lot more). So, whether it's a board with standard 4mm switches, or 1.5mm scissor switches, that relative difficulty remains. Rather than specifically try to work out distances involved down to the keypress distance, make it substantially more of penalty to go to the lower row than the upper one, and similarly follow that for other areas that will tend to need more work to reach. IoW, tweak the effort grid, and weighting of results like alternation, rolls, etc., rather than try to get the analyzer or optimizer to go too low level.

With keyboards gravitating towards thinner designs (shorter key travel distances) it is important to understand how these short travel keyboards may affect typing performance, typing forces and operator comfort. Using 15 subjects (7 males, 8 females), we wanted to determine whether there were differences in typing performance when computer operators typed on three keyboards with the same activation force (0.6 N) but with different key travel distances (2.0mm, 2.5mm and 4.0mm). During a 15 minute typing session on each keyboard, typing performance (speed and accuracy), typing forces and perceived fatigue ratings were measured. There were no differences in typing speed (p = 0.39), typing accuracy (p = 0.33) or keystroke durations (p = 0.15) across the three keyboards. However, typing force differences were measured (p < 0.003) with the longest travel keyboard (4.0mm) having higher mean and peak forces compared to the shorter travel keyboards (2.0 and 2.5 mm). These findings indicate that there is no apparent detriment in physical exposure or typing performance when using shorter travel keyboards.

Abstract from http://journals.sagepub.com/doi/pdf/10.1177/1071181311551198

And this:

Keyboards with shorter key travel are becoming widespread yet it is unknown whether there are any biomechanical differences in the typing forces when using these keyboards. If one keyboard promotes typing with more force, this may increase the risk for developing an upper extremity disorder. A total of 17 subjects typed on two short travel keyboards (<2.5 mm) and one long travel keyboard (4.0 mm). The magnitude and angle of the typing forces were measured in the x-, y- and z-axes using a thin, three-dimensional force platform. The aim of this study was to determine whether there were differences between the short and long travel keyboards in the magnitude and direction of the typing forces and the keystroke durations. In addition we wanted to determine whether there were typing force differences between key rows, hands, and fingers. Keyboards with shorter key travel resulted in less extreme angles and smaller magnitudes for forces applied in the x and y directions. Shorter travel distances were associated with smaller peak and mean vector sum forces and shorter keystroke durations. Although these results indicate keyboards with shorter key travel affect typing biomechanics, it is uncertain whether the small differences in keystroke durations and applied typing forces are physiologically meaningful and would reduce a computer user’s risk for developing an upper extremity musculoskeletal disorder.

http://journals.sagepub.com/doi/pdf/10.1177/1541931214581302