Your typing skills are totally erased: what would you learn?

Pick your new native layout.

QWERTY
36
53%
AZERTY / QWERTZ
1
1%
Dvorak
7
10%
Colemak
15
22%
Workman / Norman
2
3%
Arensito / MTGAP
2
3%
Malt / RSTHD
3
4%
Stenography
2
3%
 
Total votes: 68

squizzler

02 May 2020, 11:44

Imagine one day your typing skills were taken from you and you had to learn from scratch. Maybe there is some accident or trauma in which you are totally fine except that you have no memory on how to type - your muscle memory is gone and you cannot even remember anymore where about all the letters are. You will have to learn typing from scratch.

Which layout would you choose to learn on? Would you go with Qwerty, or would you choose an optimised one? Or something in-between? Here is your chance to choose what you would have learnt but could not justify the disruption of changing your habits. I have tried to categorise the choices in order from QWERTY, via regional typewriter configurations to Dvorak, then modern layouts for traditional staggered row keyboards such as Colemak, to Malt and RSTHD which work best on split ergonomic keyboards. Finally, if you are really hardcore you can choose stenography, but beware, this poll assumes you will learn typing from scratch and that your learning ability is unchanged, I am not offering to wave a magic wand which swaps your competence at QWERTY for an equivalent aptitude with stenotype.

You may assume that voice assistants and other 'accessibility' options continue their development and remain an alternative to some forms of typing input.

squizzler

02 May 2020, 13:46

I edited live poll to add Arensito and MTGAP which deleted previous poll results. I'm sorry, I did not know that would happen!

I have paired similar layouts in the poll to keep the number of choices to manageable levels. Let me know if that seems wrong - or if there are obvious omissions - and I can always run another poll.

User avatar
ddrfraser1

02 May 2020, 16:11

If I lost my memory I’d rewatch Star Wars

User avatar
Chyros

02 May 2020, 18:34

QWERTY. That way I can use the most keyboards. Also I doubt another layout would give me any benefit whatsoever.

User avatar
depletedvespene

02 May 2020, 19:02

QWERTY, buuuuuuuut...

Generally speaking, that only covers the letters A-Z, numbers and two symbols (comma and dot). There are many national layouts that are QWERTY but differ greatly in the placement of symbols and diacritics (English (US, UK, others), Spanish (three), Portuguese (two), Italian, Danish, Norwegian, Swedish/Finnish, etcetera. QWERTZ is generally associated with the German national layout, but some other national layouts use it as well. AZERTY is pretty much only the domain of French in France and Belgium (with the differences between those being minor).

Dvorak? Colemak? With the notable exception of BÉPO, all those alternative letter arrangements were made with English in mind and not much research exists about their suitability for other languages (even though "versions" of Dvorak exist for, say, Spanish, but those tend to not actually move the basic A-Z letters around).

So, would it be enough to say "QWERTY vs. QWERTZ vs. DVORAK", or should the distinction between this component and a full national layout be made?

squizzler

02 May 2020, 19:46

depletedvespene wrote:
02 May 2020, 19:02
QWERTY, buuuuuuuut...

Generally speaking, that only covers the letters A-Z, numbers and two symbols (comma and dot). There are many national layouts that are QWERTY but differ greatly in the placement of symbols and diacritics

<snip!>

So, would it be enough to say "QWERTY vs. QWERTZ vs. DVORAK", or should the distinction between this component and a full national layout be made?
You make a good point and one which I did actually consider making a disclaimer that I consider only Latin layouts, but dismissed doing so on the basis that this forum is English speaking (or more specifically, English typing!). You are quite right that other languages need access to various other elements those of us in the Anglosphere don't consider. Please accept my apologies for this!

For the purpose of this survey, I personally think it is quite sufficient to only distinguish between QWERTY as a whole rather than national variations within that.

User avatar
depletedvespene

02 May 2020, 19:56

squizzler wrote:
02 May 2020, 19:46
depletedvespene wrote:
02 May 2020, 19:02
QWERTY, buuuuuuuut...

Generally speaking, that only covers the letters A-Z, numbers and two symbols (comma and dot). There are many national layouts that are QWERTY but differ greatly in the placement of symbols and diacritics

<snip!>

So, would it be enough to say "QWERTY vs. QWERTZ vs. DVORAK", or should the distinction between this component and a full national layout be made?
You make a good point and one which I did actually consider making a disclaimer that I consider only Latin layouts, but dismissed doing so on the basis that this forum is English speaking (or more specifically, English typing!). You are quite right that other languages need access to various other elements those of us in the Anglosphere don't consider. Please accept my apologies for this!

For the purpose of this survey, I personally think it is quite sufficient to only distinguish between QWERTY as a whole rather than national variations within that.
Oh, well, as long as we are restricting ourselves to English... my answer would be NONE OF THE ABOVE! :D

Knowing myself, I'd gather up a list of letters (A-Z), numbers (0..9, not 1..9 & 0) and symbols ( ,;.:+-*/\| () [] {} <> «» ¡! ¿? '"#$€ — – ¹²³ⁿ =≠ etc.) that I feel that need to be present on the (English) keyboard, and jumble them until I was satisfied with the result as good enough for me (and, by extension, everyone else in the (English speaking) world). Y'all know the great thing about standards, right? :D

edlee

02 May 2020, 20:21

I would create my own layout and begin with a sequence of keys on the home row spelling out the word, "the", from left to right, because, "the", is the most common word in the English language.

kelvinhall05

03 May 2020, 01:00

Man, there would be so many I would love to learn...I really want to learn Steno at some point, and I would like to know if there is actually a speed benefit to dvorak, but considering that like 90% of keyboards are just qwerty, I think I'd stick with that.

zzxx53

03 May 2020, 06:57

QWERTY so that I can still use most keyboards around me.
Stenography doesn't help at all when most of what you type isn't natural language text, which is the case for many software developers.

User avatar
Sup

03 May 2020, 08:18

I would switch to colemak or Dvorak with a ortholinear board for good measure

squizzler

03 May 2020, 12:04

Thanks for all the votes cast so far!
Chyros wrote:
02 May 2020, 18:34
QWERTY. That way I can use the most keyboards. Also I doubt another layout would give me any benefit whatsoever.
I cannot argue with that logic, as a keyboard collector and reviewer it would be very hard to justify anything else!
Sup wrote:
03 May 2020, 08:18
I would switch to colemak or Dvorak with a ortholinear board for good measure
These are the layouts that I originally thought would dominate the poll. Whilst newbies don't know that there are alternative layouts to pick, we know of the existence layouts widely regarded as more efficient than Qwerty. We are also aware that the better known layouts are well supported, and most of us run keyboards with discrete switches which allow the caps to be detached and rearranged or swapped with correctly sculpted sets for common alternatives.

If you were learning typing on an ortholinear board, maybe you would benefit from a Malt style layout? Those which put the 'E' on the thumb such as RSTHD are consistently scored more highly by analyser programmes than the best of the layouts that fit on legacy staggered row hardware.

User avatar
webwit
Wild Duck

03 May 2020, 12:53

I actually rebooted in 2007.

Before I was as lame as Chyros. :twisted: Never officially learned to touch type and I used some half-assed hunt and peck method of typing. I thought, I'm going to type for another couple of decades with any luck, let's try something better.

So first I learned to touchtype and simultaneously switched to Colemak while I was at it. And switched to HHKB. This proved to be a winning move because 1) Colemak is so much better, 2) Blind typing allows me to use any keyboard whatever the labels say and requires so much less effort after you got the muscle memory, 3) Because it's such a different mode I'm in, I can still use QWERTY when I need to, even after all this time.

Then I went all the way, and ditched traditional keyboards over the total reboot which is the Datahand (using Colemak as well).
Spoiler:
Image
I typed on it for a couple of years and it makes normal keyboards primitive. However, since they went out of business, the build quality isn't all that, and portability issues, I switched back to the HHKB/Colemak. I didn't want to become totally dependent on a dead keyboard. Using that since then, until recently when I found my end game in the Model F62 HHKB with Colemak.

I really love that form factor and moreover, it's closest to the Datahand when it comes to hand travel. There's zero with the Datahand, but the HHKB layout with Colemak layout is pretty good as well. It is not often I still need to use Qwerty, because I'm almost never typing on someone else's keyboard, but god these big Qwerty keyboards look so silly now with your hands moving all over the place.

I didn't regret my investment for a single moment, it's like when you need to walk for a profession and I finally got some good professional boots instead of having walked on flip-flops for so long like an idiot.

davkol

03 May 2020, 13:46

zzxx53 wrote:
03 May 2020, 06:57
Stenography doesn't help at all when most of what you type isn't natural language text, which is the case for many software developers.
Why would you think that?

You can have a shorthand for any (non-random) corpus, and coming up with one for a severely restricted formal language is likely easier than dealing with natural languages. People in the OpenSteno/Plover project have come up with example dictionaries for such work, and it seems someone has actually used it in practice.

Writing in multiple natural languages is a much bigger challenge, because each language typically requires a distinct "theory".
depletedvespene wrote:
02 May 2020, 19:02
Dvorak? Colemak? With the notable exception of BÉPO, all those alternative letter arrangements were made with English in mind and not much research exists about their suitability for other languages (even though "versions" of Dvorak exist for, say, Spanish, but those tend to not actually move the basic A-Z letters around).
On the other hand, the commonly used [Indo-European] languages are close enough, assuming the common alphabetic writing system (esp. w/ the Latin alphabet) and standard keyboard input methods.

Even though n-gram frequencies may be slightly different and some orthographies make use of accents and what not, the fundamental typing rhythm translates pretty well: alternating vowels and consonants in syllables.

Thus, typing in Dvorak involves a lot of hand alternation, Maltron/Colemak/MTGAP/… are more on the "rolling" side, and QWERTY (& its various localizations) is an inconsistent mess regardless of the specific language.

source: I write a fair bit in Czech (which is substantially different from English in terms of grammar and orthography), and I have run the numbers.
edlee wrote:
02 May 2020, 20:21
I would create my own layout and begin with a sequence of keys on the home row spelling out the word, "the", from left to right, because, "the", is the most common word in the English language.
That completely misses the point. Even as the most common trigram in English, it's still less than 2 % of actual text, while you would be likely sacrificing the experience of typing other n-grams.

However, I have seen someone suggest a distinct key for 'th' (or even coming back with ð or þ).
squizzler wrote:
03 May 2020, 12:04
If you were learning typing on an ortholinear board, maybe you would benefit from a Malt style layout? Those which put the 'E' on the thumb such as RSTHD are consistently scored more highly by analyser programmes than the best of the layouts that fit on legacy staggered row hardware.
What does it have to do with the layout being "ortholinear"?

You can have multiple thumb keys on a traditional asymmetrically row-staggered keyboard.

Placing a common symbol under a thumb has two benefits:
  • spreading load potentially more evenly (which positively impacts writing in various languages as well)
  • L. Malt also claimed that separating vowels reduced error rates (a factor ignored by all the new machine-generated layouts, such as the RSTHD—thus inappropriately—grouped with the Maltron layouts)
On the other hand, "ortholinear" keyboards have straight rows just like the standard keyboards, whereas L. Malt explicitly referenced staggered columns as means to make little fingers first-class citizens on the keyboard, and showed in one of her papers that keystroke transitions involving little fingers were indeed significantly faster (less slow) on a columnar layout than otherwise.

This leads to another point…
zzxx53 wrote:
03 May 2020, 06:57
QWERTY so that I can still use most keyboards around me.
Most keyboards aren't technically QWERTY, but rather the software keymap is set up as QWERTY—or something else in other environments, where people speak a different language.

Frankly, this is my main concern. While it's easy to switch keymaps in software, the hardware isn't going anywhere, unfortunately. That's why I'm thinking of building my own cyberdeck instead of importing at least Japanese keyboard modules for each and every new laptop I might end up using.

squizzler

03 May 2020, 16:13

davkol wrote:
03 May 2020, 13:46
squizzler wrote:
03 May 2020, 12:04
If you were learning typing on an ortholinear board, maybe you would benefit from a Malt style layout? Those which put the 'E' on the thumb such as RSTHD are consistently scored more highly by analyser programmes than the best of the layouts that fit on legacy staggered row hardware.
What does it have to do with the layout being "ortholinear"?

You can have multiple thumb keys on a traditional asymmetrically row-staggered keyboard.

Placing a common symbol under a thumb has two benefits:
  • spreading load potentially more evenly (which positively impacts writing in various languages as well)
  • L. Malt also claimed that separating vowels reduced error rates (a factor ignored by all the new machine-generated layouts, such as the RSTHD—thus inappropriately—grouped with the Maltron layouts)
On the other hand, "ortholinear" keyboards have straight rows just like the standard keyboards, whereas L. Malt explicitly referenced staggered columns as means to make little fingers first-class citizens on the keyboard, and showed in one of her papers that keystroke transitions involving little fingers were indeed significantly faster (less slow) on a columnar layout than otherwise.
You caught out my schoolboy error - I was of course confusing "ortholinear" with what might properly be described as "Maltron inspired columnar staggered ergonomic board with thumb clusters". I want to tell all *Dox owners that this style of board and the Malt layout were designed as an integrated system, and they owe it to themselves to use a layout whose 'E' or other vital character is on the thumb!

I thank you also for the rest of your post which leaves me with a lot of food for thought - It is good to meet somebody here who is apparently a Malt aficionado! I actually made my first jump from Qwerty with Malt trial. Of course I kept reading and discovered RSTHD. I was receptive to the critique of malt offered by the inventor of that layout, particularly the reference to the 'pinky twist' from the location of the letter 'L' in the top right. I lived recently in Wales and if you made an optimum layout for Welsh I assure you the 'L' would be on the home row. :D


I can see why the consider that vowel placement means that RSTHD is not strictly not built on Malt principles, although its designer is quite open in Malt having been an inspiration. Given that some similar layouts are combined into one voting choice for the purposes of this survey, I felt the all layouts with 'E' on the thumb would be a minor category even combined, although maybe it will get enough votes to justify future research!. Coming from a Malt trial the placement of vowels initially struck me as 'not looking right' to see them all (except 'E') clumped, although Colemak and Workman both have a similar clump of four vowels on the right hand side, suggesting that is probably close to optimum. RSTHD requires one finger to cover two vowels which does not seem an excessive workload.
edlee wrote:
02 May 2020, 20:21
I would create my own layout and begin with a sequence of keys on the home row spelling out the word, "the", from left to right
There is no need to create your own layout since RSTHD effectively does just that: as you might deduce from its name T and H are adjacent on the left hand home row and E is activated by the thumb. A lovely inwards roll with the three adjacent strongest digits!

stevep

03 May 2020, 19:00

I don't understand why the poll has some seemingly unrelated layouts (e.g. Norman and Workman) grouped together as though they were the same thing. Norman and Workman are very different. It would make more sense to either separate them, or just list the 3 or 4 more likely options and then have "other".

I'm also amazed at how many people pick Qwerty. It's one thing to not want to go through the change if you've used it all your life, but starting with a clean slate and yet *still* choosing by far the worst option seems utterly absurd to me.

kelvinhall05

03 May 2020, 19:43

stevep wrote:
03 May 2020, 19:00
I don't understand why the poll has some seemingly unrelated layouts (e.g. Norman and Workman) grouped together as though they were the same thing. Norman and Workman are very different. It would make more sense to either separate them, or just list the 3 or 4 more likely options and then have "other".

I'm also amazed at how many people pick Qwerty. It's one thing to not want to go through the change if you've used it all your life, but starting with a clean slate and yet *still* choosing by far the worst option seems utterly absurd to me.
I think that basically all of us here have the mindset of "qwerty kinda sucks but is by FAR the most common and it is a bit of a pita to use others on a day-to-day basis regardless of if we had to learn it or not". Know what I mean?


I agree, if all of those layouts had equal presence in society then I don't think anyone here would pick qwerty...I would probably pick something like dvorak and/or steno.

davkol

03 May 2020, 21:21

stevep wrote:
03 May 2020, 19:00
I don't understand why the poll has some seemingly unrelated layouts (e.g. Norman and Workman) grouped together as though they were the same thing. Norman and Workman are very different. It would make more sense to either separate them, or just list the 3 or 4 more likely options and then have "other".
That's a good point, because…
squizzler wrote:
03 May 2020, 16:13
Of course I kept reading and discovered RSTHD. I was receptive to the critique of malt offered by the inventor of that layout, particularly the reference to the 'pinky twist' from the location of the letter 'L' in the top right. I lived recently in Wales and if you made an optimum layout for Welsh I assure you the 'L' would be on the home row. :D

I can see why the consider that vowel placement means that RSTHD is not strictly not built on Malt principles, although its designer is quite open in Malt having been an inspiration. Given that some similar layouts are combined into one voting choice for the purposes of this survey, I felt the all layouts with 'E' on the thumb would be a minor category even combined, although maybe it will get enough votes to justify future research!. Coming from a Malt trial the placement of vowels initially struck me as 'not looking right' to see them all (except 'E') clumped, although Colemak and Workman both have a similar clump of four vowels on the right hand side, suggesting that is probably close to optimum. RSTHD requires one finger to cover two vowels which does not seem an excessive workload.
The only thing this 'RSTHD' layout has to do with Maltron is the 'e' placement, any basic touch-typing optimization aside. So does, for example, the AdNW variant for ergonomic keyboards. The remaining design principles are apparently ignored (if the author ever cared enough to learn about them in the first place).

How many people even use those layouts (RSTHD, BEAKL, MTGAP,…)? A dozen?
Are they based on any serious research, or just arbitrary toy projects? From the linked blog: "I tried … swapping keys around until I saw something I liked while trying to balance out hand and finger usage and eyeballing digraph comfortableness by intuition." :roll: There's no justification in the README either, only normative claims that "this is awkward" and "that's slow".
Why should anyone care? There are plenty of such permutations. (see also quadratic assignment problem)
stevep wrote:
03 May 2020, 19:00
I'm also amazed at how many people pick Qwerty. It's one thing to not want to go through the change if you've used it all your life, but starting with a clean slate and yet *still* choosing by far the worst option seems utterly absurd to me.
Keyboard shortcuts are often designed for QWERTY, thus QWERTY is 'optimal' for that use case in a way.

Then, it's not "by far the worst option". It's not the worst layout for typing either, even though it does pre-date touch typing centered around the home row and therefore lacks any overarching flow.

JBert

04 May 2020, 02:29

I was constantly switching between AZERTY, QWERTY and QWERTY (US International ). My layout at home was always QWERTY, but at work I sometimes needed French characters.

Switching gradually drove me crazy: it meant inadvertently pressing CTRL-W when I meant CTRL-Z or CTRL-Q when I needed CTRL-A.

Since I wanted to do something about my horrible hunt-and-peck-trigrams technique I started over in Colemak, and once I got fast enough (after one month at least) I simply started looking into permanently using it.

Colemak on Windows (and Linux too I believe) come with a pretty useful AltGr layer out of the box. Nearly any character used in West-European languages is there, even if a few are using a dead key sequence.

I would imagine that other layouts might not have had the foresight of including a (standardized) international character layer. I could imagine that Dvorak's is different in all its variants because it was originally designed for writing English on a typewriter.

edlee

04 May 2020, 06:56

No single keyboard layout is optimal for everyone's uses. If you get something like Soarer's converter or Hasu's USB to USB converter, then you can develop your own keyboard layout and carry it around with you wherever you go. I recently added macros to my Soarer configuration file to produce blocks of Visual Basic code, "if then end if", "if then else end if", "while wend", "select case end select", "for next", "do loop". Once you have the ability to program macro keys, you are no longer limited to one character per key.

squizzler

04 May 2020, 09:41

stevep wrote:
03 May 2020, 19:00
I don't understand why the poll has some seemingly unrelated layouts (e.g. Norman and Workman) grouped together as though they were the same thing. Norman and Workman are very different. It would make more sense to either separate them, or just list the 3 or 4 more likely options and then have "other".
I tried to reduce options to make the poll simpler, so the legacy European layouts (AZERTY, QWERTZ) are lumped together. Norman and Workman are more recent than Colemak, and probably inspired by interest in the latter, so it made sense to combine them.
davkol wrote:
03 May 2020, 21:21
That's a good point, because…
squizzler wrote:
03 May 2020, 16:13
Of course I kept reading and discovered RSTHD. I was receptive to the critique of malt offered by the inventor of that layout, particularly the reference to the 'pinky twist' from the location of the letter 'L' in the top right. I lived recently in Wales and if you made an optimum layout for Welsh I assure you the 'L' would be on the home row. :D

I can see why the consider that vowel placement means that RSTHD is not strictly not built on Malt principles, although its designer is quite open in Malt having been an inspiration. Given that some similar layouts are combined into one voting choice for the purposes of this survey, I felt the all layouts with 'E' on the thumb would be a minor category even combined, although maybe it will get enough votes to justify future research!. Coming from a Malt trial the placement of vowels initially struck me as 'not looking right' to see them all (except 'E') clumped, although Colemak and Workman both have a similar clump of four vowels on the right hand side, suggesting that is probably close to optimum. RSTHD requires one finger to cover two vowels which does not seem an excessive workload.
The only thing this 'RSTHD' layout has to do with Maltron is the 'e' placement, any basic touch-typing optimization aside. So does, for example, the AdNW variant for ergonomic keyboards. The remaining design principles are apparently ignored (if the author ever cared enough to learn about them in the first place).

How many people even use those layouts (RSTHD, BEAKL, MTGAP,…)? A dozen?
Are they based on any serious research, or just arbitrary toy projects? From the linked blog: "I tried … swapping keys around until I saw something I liked while trying to balance out hand and finger usage and eyeballing digraph comfortableness by intuition." :roll: There's no justification in the README either, only normative claims that "this is awkward" and "that's slow".
Why should anyone care? There are plenty of such permutations. (see also quadratic assignment problem)
I agree that these are not massively popular layouts at the time of writing, but the same could be said about your favoured Malt layout. The introduction of innumerable *dox variations, Corne, Keyboardio, etc, have lowered the barrier to entry to layouts that assign a character to the thumb and there are possibly now more people using Malt on non Malt boards than on the real McCoy.

Your use selective quoting to make it seem that the RSTHD layout is arbitrarily designed. The author goes on to explain he then went on to design an application to sort through the different configurations and, assuming that he based his scoring on correct assumptions, the winning layout would be the best layout, even if it were created by monkeys!

I share your cynicism about computer models for typing that spit out all sorts of random layouts and also believe in a human factor. I have not the skill to verify his findings, but intuition tells me the layout is a good 'un. Features such as the configuration of the vowels (a clump of three along the right of the home row and the letter 'U' on its own above) are found in Colemak and Workman.

I have found this brief paper on the Maltron website. It must be remembered that since Malt took a holistic view of keyboard design and letter layout, and packed so many innovations into her design, some of those assumptions and design choices were bound to prove less important than others. The vowel placement to prevent "neural confusion" does not seem be a major consideration in subsequent alternative layouts mentioned above. The accuracy of typing is important but perhaps the prevalence of autocorrect these days might reduces its importance?

I feel that the Malt layout (based on all the same design principles) would look different if she had access to the resources available to what you dismiss as the hobbyist of today. Not that Maltron is a bad layout even in the very narrow definition of typing efficiency you critique. The scoring table ranks Malt more highly than Colemak et.al.

Image

Thus I think it is important not to consider the Malt and RSTHD layouts to be locked in a zero-sum game. They increase the utility of compatible ergonomic boards and thus the success of one is likely to lead to greater success for the other.
stevep wrote:
03 May 2020, 19:00
I'm also amazed at how many people pick Qwerty. It's one thing to not want to go through the change if you've used it all your life, but starting with a clean slate and yet *still* choosing by far the worst option seems utterly absurd to me.
Keyboard shortcuts are often designed for QWERTY, thus QWERTY is 'optimal' for that use case in a way.

Then, it's not "by far the worst option". It's not the worst layout for typing either, even though it does pre-date touch typing centered around the home row and therefore lacks any overarching flow.
This is an interesting point, although I understand the more refined alternative layouts are mindful of the placement of common shortcuts such as 'copy' and 'paste'. At any rate, with firmware driven keyboards it is possible to assign shortcuts to practically any set of keystrokes you choose.

davkol

04 May 2020, 10:39

squizzler wrote:
04 May 2020, 09:41
Your use selective quoting to make it seem that the RSTHD layout is arbitrarily designed. The author goes on to explain he then went on to design an application to sort through the different configurations and, assuming that he based his scoring on correct assumptions, the winning layout would be the best layout, even if it were created by monkeys!
That's precisely the point.

He doesn't justify his assumptions anywhere, unlike Dvorak (et al.) who published a whole book about typing… or Malt who also released at least a few papers with rudimentary experiments and had to go through the process of patenting the layout.

If you want another example, here's a layout based on Claude Marsan's work from the 1970s, but from a ~2003 paper that used an ant colony optimization algorithm; the authors asked a couple of professional ergonomists to reach consensus on optimization criteria… far from great, at least somewhat qualified, though. (The layout looks weird.)

Image
squizzler wrote:
04 May 2020, 09:41
The scoring table ranks Malt more highly than Colemak et.al.
I can come up with any number of such arbitrary tables.
squizzler wrote:
04 May 2020, 09:41
The vowel placement to prevent "neural confusion" does not seem be a major consideration in subsequent alternative layouts mentioned above. The accuracy of typing is important but perhaps the prevalence of autocorrect these days might reduces its importance?
I haven't seen any evidence that it has even occurred to those people that error rates may be a concern.

That being said, I would also like to see hard data to test Malt's (and Dvorak's) hypotheses.
squizzler wrote:
04 May 2020, 09:41
I feel that the Malt layout (based on all the same design principles) would look different if she had access to the resources available to what you dismiss as the hobbyist of today. Not that Maltron is a bad layout even in the very narrow definition of typing efficiency you critique.
The current hobbyists only have faster computers, while apparently putting no effort into at least looking up medical research (or even textbooks), caring about cognition or doing reproducible experiments.

Zero, zilch, zip, nada, nothing.

squizzler

05 May 2020, 14:32

davkol wrote:
04 May 2020, 10:39
squizzler wrote:
04 May 2020, 09:41
Your use selective quoting to make it seem that the RSTHD layout is arbitrarily designed. The author goes on to explain he then went on to design an application to sort through the different configurations and, assuming that he based his scoring on correct assumptions, the winning layout would be the best layout, even if it were created by monkeys!
That's precisely the point.

He doesn't justify his assumptions anywhere, unlike Dvorak (et al.) who published a whole book about typing… or Malt who also released at least a few papers with rudimentary experiments and had to go through the process of patenting the layout.

If you want another example, here's a layout based on Claude Marsan's work from the 1970s, but from a ~2003 paper that used an ant colony optimization algorithm; the authors asked a couple of professional ergonomists to reach consensus on optimization criteria… far from great, at least somewhat qualified, though. (The layout looks weird.)

Image
Good heavens weird is an understatement for that last arrangement! It does seem to confirm that a design developed completely from first principles by qualified scientists will not be always better!

The inventor of RSTHD has open-sourced his programme, should any of us be qualified to dive in. I am not and I have no answer to your objective line of question. But the author's 'journey' to reach his layout accorded to my own experience and stage of my keyboard journey as a seasoned-but-not-yet-committed Malt user. Sorry to be so wishy-washy :D . I think the author is perhaps a little modest about his achievement. He has created something for his own use, been good enough to publish it, and the comments suggest that others have also adopted it. What it really needs is a community of users to continue its development and advocacy.
squizzler wrote:
04 May 2020, 09:41
The scoring table ranks Malt more highly than Colemak et.al.
I can come up with any number of such arbitrary tables.
I think the table does seem to confirm what one might feel would be the case: any scientifically designed layout like Dvorak represents a considerable leap over QWERTY but there is a diminishing return associated with more refined layouts. The more popular layout analyser seems to plot a much more straight-line (and thus implausible) relationship from Qwerty - Dvorak - Colemak - Random layout-de-jour.

The column names are instructive too: 'Finger Pinky Twist' and 'Left Crunch' (whatever that is!) imply the model is based more on biology than some, which I suspect treat finger movement as cartesian movement through the X and Y axis.
squizzler wrote:
04 May 2020, 09:41
The vowel placement to prevent "neural confusion" does not seem be a major consideration in subsequent alternative layouts mentioned above. The accuracy of typing is important but perhaps the prevalence of autocorrect these days might reduces its importance?
I haven't seen any evidence that it has even occurred to those people that error rates may be a concern.

That being said, I would also like to see hard data to test Malt's (and Dvorak's) hypotheses.
squizzler wrote:
04 May 2020, 09:41
I feel that the Malt layout (based on all the same design principles) would look different if she had access to the resources available to what you dismiss as the hobbyist of today. Not that Maltron is a bad layout even in the very narrow definition of typing efficiency you critique.
The current hobbyists only have faster computers, while apparently putting no effort into at least looking up medical research (or even textbooks), caring about cognition or doing reproducible experiments.

Zero, zilch, zip, nada, nothing.
Today's layout hobbyists not only have faster computers (and the tools to harness this power to assist layout design) but there is a community of such people for whom a lot of this knowledge has become internalised - what Brian Eno called 'scenius'. For example, as a hobby rider, I could buy a bicycle frame and all the component parts and build a bespoke bicycle in my garage that would be more comfortable and efficient than a top-of-the-line bike from Raleigh in the 1970's: then a serious company with a large factory employing skilled people in drawing offices and research laboratories. Raleigh, with all its resources, was producing the 'Chopper', an overweight, ill handling monstrosity of a thing.

The amateur bicycle club racer of today will probably choose more healthy food and more effective exercising regimes than a professional rider of the 1970s with full-time team support, simply because the amateur racing scene is immersed in better knowledge and has more sophisticated tools (Strava, home turbo trainers and onboard power monitors) than would be available to the medical professional in the 1970s with access to the latest academic papers. I suggest it is probably the same with today's amateur keyboard layout crafters.

This has been a very illuminating debate which I feel might justify being assigned its own thread on 'Malt versus other layouts with thumb placement of vowels'. There is one thing which I think we would both agree on and that is Malt layout deserved greater success than it has hitherto enjoyed.

davkol

05 May 2020, 15:54

squizzler wrote:
05 May 2020, 14:32
The inventor of RSTHD has open-sourced his programme, should any of us be qualified to dive in.
There is about a dozen of such programs publicly available, and a semi-competent coder can make a new one from scratch over a weekend. Been there, done that.

That's not the problem. The problem is figuring out the parameters for running the code: Garbage In – Garbage Out

You seem to be consistently ignoring this part.
squizzler wrote:
05 May 2020, 14:32
The column names are instructive too: 'Finger Pinky Twist' and 'Left Crunch' (whatever that is!) imply the model is based more on biology than some, which I suspect treat finger movement as cartesian movement through the X and Y axis.
This points out another failure of these models: they assume a simplistic abstraction of the typing technique.

However, actual experienced typists don't type that way. For example, the 'rt' bigram implies using the index finger twice in a row based on the rigid touch-typing scheme, but a real person might just as well shift their hand one column and use both middle and index finger.

Touch typing is a quite dynamic process, if you look at the typists' hands. People don't strike individual keys in isolation; the movement is adjusted based on multiple previous and in anticipation of multiple following keystrokes. It's documented in the book Cognitive Aspects of Skilled Typewriting IIRC.

IOWs the typist might just as well twist or shift their whole hand/arm to avoid an awkward motion or offload some effort to stronger muscles.

Such workarounds further reduce the differences between layouts in terms of individual metrics. Some metrics do look bad on paper indeed, but it's less of an issue with practice. You might point out that the typist has to learn the more complex motions first, and fair enough. Do you see hobbyists design layouts to be easy to learn or require as few distinct motions as possible, though? I don't, unless you count using simplified models as exactly that.
squizzler wrote:
05 May 2020, 14:32
Today's layout hobbyists not only have faster computers (and the tools to harness this power to assist layout design) but there is a community of such people for whom a lot of this knowledge has become internalised
I call bullshit.

When you talk about bikers, you omit the part about active research and experiments; hardly anyone is doing that with keyboards (certainly not in the forums, past posting 'wpm/accuracy' from simplistic tests) and hardly anyone reads about research that did that decades ago. It wouldn't even have to be keyboard-specific: I'm sure there are, for example, plenty of studies on playing musical instruments that may translate into typing pretty well.

Malt referenced, e.g., early research on neural response latency that, she hypothesized, would make Dvorak layout slower than Maltron due to more frequent hand alternation. Where are all the forum posts dealing with this hypothesis? I has the potential to make Dvorak-like layouts decidedly obsolete.

User avatar
robo

05 May 2020, 22:09

Just a data point of 1, but here's my totally scientific reason why I'd just pick QWERTY:

- When I was about 12, I learned to touch type in QWERTY.
- When I was 18, I learned about Dvorak, thought it was cool, and taught myself to type Dvorak.
- After 9 years of exclusively typing Dvorak (and typing a lot), I got a job where it was hugely inconvenient to not be able to type QWERTY (ie. had to type on keyboards where I could not switch the layout), so I decided it would be a career enhancing move to quit Dvorak and go back to QWERTY. I wanted to know how much typing speed I'd lose, so I benchmarked my typing speed before going cold turkey with QWERTY. Took me about a week to be comfortable. After about a month back on QWERTY, I benchmarked my typing speed again.

It was IDENTICAL. Literally exactly the same. I think it was about 85WPM on both.

So, I concluded that more 'efficient' layouts are not worth the hassle of being helpless at anyone else's keyboard (without fiddling with switching layouts etc).

User avatar
Chyros

05 May 2020, 22:30

robo wrote:
05 May 2020, 22:09
Just a data point of 1, but here's my totally scientific reason why I'd just pick QWERTY:

- When I was about 12, I learned to touch type in QWERTY.
- When I was 18, I learned about Dvorak, thought it was cool, and taught myself to type Dvorak.
- After 9 years of exclusively typing Dvorak (and typing a lot), I got a job where it was hugely inconvenient to not be able to type QWERTY (ie. had to type on keyboards where I could not switch the layout), so I decided it would be a career enhancing move to quit Dvorak and go back to QWERTY. I wanted to know how much typing speed I'd lose, so I benchmarked my typing speed before going cold turkey with QWERTY. Took me about a week to be comfortable. After about a month back on QWERTY, I benchmarked my typing speed again.

It was IDENTICAL. Literally exactly the same. I think it was about 85WPM on both.

So, I concluded that more 'efficient' layouts are not worth the hassle of being helpless at anyone else's keyboard (without fiddling with switching layouts etc).
This is not surprising, really. Many studies have been done on the efficiency of QWERTY versus several self-proclaimed "efficient" layouts like DVORAK, and there is still no clear conclusion as to whether it works or not; no conclusive evidence has been shown for it to provide for faster typing speed. Between the large switching barrier, incompatibility with the vast majority of keyboards in the field, and the lack of clear efficiency benefits, it's no wonder the QWERTY layout has been holding all this time.

kelvinhall05

05 May 2020, 22:32

Chyros wrote:
05 May 2020, 22:30
robo wrote:
05 May 2020, 22:09
Just a data point of 1, but here's my totally scientific reason why I'd just pick QWERTY:

- When I was about 12, I learned to touch type in QWERTY.
- When I was 18, I learned about Dvorak, thought it was cool, and taught myself to type Dvorak.
- After 9 years of exclusively typing Dvorak (and typing a lot), I got a job where it was hugely inconvenient to not be able to type QWERTY (ie. had to type on keyboards where I could not switch the layout), so I decided it would be a career enhancing move to quit Dvorak and go back to QWERTY. I wanted to know how much typing speed I'd lose, so I benchmarked my typing speed before going cold turkey with QWERTY. Took me about a week to be comfortable. After about a month back on QWERTY, I benchmarked my typing speed again.

It was IDENTICAL. Literally exactly the same. I think it was about 85WPM on both.

So, I concluded that more 'efficient' layouts are not worth the hassle of being helpless at anyone else's keyboard (without fiddling with switching layouts etc).
This is not surprising, really. Many studies have been done on the efficiency of QWERTY versus several self-proclaimed "efficient" layouts like DVORAK, and there is still no clear conclusion as to whether it works or not; no conclusive evidence has been shown for it to provide for faster typing speed. Between the large switching barrier, incompatibility with the vast majority of keyboards in the field, and the lack of clear efficiency benefits, it's no wonder the QWERTY layout has been holding all this time.
Haha, iirc the only person who claimed dvorak was faster was the dude who invented it :lol:

User avatar
webwit
Wild Duck

05 May 2020, 23:07

robo wrote:
05 May 2020, 22:09
It was IDENTICAL. Literally exactly the same. I think it was about 85WPM on both.
That's roughly my speed. I think if you're operating at that speed, unlike a secretary keeping up with fast speech but not thinking about what is being entered, or someone really linguistically intelligent and thinking really fast in language, the limiting factor is not the layout but the brain. When I'm discussing or coding, it's not in 130WPM.

Then I think you're looking for the wrong thing in superior layouts, that is breaking the speed record.

(Of course our limited speed could mean you and/or I could be a bit of a dolt, but in our defense, Einstein was a self-confessed slow thinker, i.e. fast intelligence does not equal best intelligence by default.)

To repeat my earlier metaphor, when you walk for a living, would you buy pro walking shoes or do it on flip-flops. The speed of walking would be the same, but how do your feet feel after a day walking on both? Qwerty is a lot more work. Get pro gear.

davkol

05 May 2020, 23:44

kelvinhall05 wrote:
05 May 2020, 22:32
Chyros wrote:
05 May 2020, 22:30
This is not surprising, really. Many studies have been done on the efficiency of QWERTY versus several self-proclaimed "efficient" layouts like DVORAK, and there is still no clear conclusion as to whether it works or not; no conclusive evidence has been shown for it to provide for faster typing speed. Between the large switching barrier, incompatibility with the vast majority of keyboards in the field, and the lack of clear efficiency benefits, it's no wonder the QWERTY layout has been holding all this time.
Haha, iirc the only person who claimed dvorak was faster was the dude who invented it :lol:
Simplified Keyboard users were overrepresented among top-ranking participants in typing competitions (while Dvorak supposedly paid for their travel from his own income and savings); the layout was actually taught to thousands of students (still a very small number compared to QWERTY) and shown to be easier to learn too.

Excerpt from Yamada's paper:

Image

edit: run OCR on that paper, here it is for search engines, accessibility etc.
DSK (and its variants) is the only keyboard among all alternatives to Qwerty before and after it which has actually been extensively tested in pedagogical environments. To name a few, tests were made at the University of Washington (Uhl & Dvorak 1933, Dvorak & Ford 1933), at the University of Minnesota (Wallace 1934), at a junior college, senior and junior high schools (Davis 1935), in the seventh grade (Merrick 1938), and recently at Carmel Middle School (Kolb 1978). In Tacoma School District, Washington, alone, 2,700 students were put through DSK typing courses in 1930's (Parkinson 1972).

The Australian Post Office undertook in 1952 an evaluation of DSK through the training of new typists by carefully observing the progress of typists on Qwerty keyboard and DS keyboard. The data of the study did not show on the surface any advantage of DSK over Qwerty. Careful analysis of the experiment revealed, however, that there was a strong psychological force of conformism in action such that the DSK group did not wish to make any further effort to outperform Qwerty group, after they accomplished as much as what Qwerty group did. Thus, another experiment was conducted in 1953, this time with incentives for better performance for each group. With such arrangement, there emerged clear evidence of the advantage of DSK over Qwerty (Australian Post Office 1953).

The results of all these tests have been well documented and for the details the readers must be referred to the references. We must be content here with the listing of bare facts on the advantage of DSK.

(a) Easier to learn: it usually takes one third of the time compared with Qwerty to get up to the same level of competence at the early stage of learning. Figure 8 shows the learning curves of U.S. high school students on DSK and on Qwerty, plotted for the best students (100% lines), for the worst of the top quarter (75 % line), and so on. The DSK students obtain after 180 hors (or, two semesters) a higher skill than the Qwerty students after 540 hours (or, six semesters).
(b) More accurate: the errors made by DSK users are approximately one half of that by Qwerty users.
(c) Faster to work with: the DSK users type 15 to 20% faster for timed copy-typing of limited durations than the Qwerty users, and 25 to 50% faster in routine production typing of every day work.
(d) Less fatiguing: all typists who have switched from Qwerty to DSK attest that DSK leaves them much less fatigued at the end of day's work.

In spite of the fact that the DSK trainees were out-numbered by Qwerty trainees perhaps by the ratio of several hundred to one, the DSK trained typists swept the field and won most of the 1st, 2nd and 3rd places in the International Commercial Schools Contest from 1933 to 1941 (Educational Test Bureau 1950).

It is interesting to note the fact that the winners used DSK was often very carefully suppressed; for example, Smith's report (1945) does not mention the fact at all.
Hisao Yamada: "A Historical Study of Typewriters and Typing Methods: from the Position of Planning Japanese Parallels", Journal of Information Processing, 2(4) (February 1980), pp. 175–202
Last edited by davkol on 14 May 2020, 11:18, edited 1 time in total.

squizzler

06 May 2020, 08:06

robo wrote:
05 May 2020, 22:09
After 9 years of exclusively typing Dvorak (and typing a lot), I got a job where it was hugely inconvenient to not be able to type QWERTY (ie. had to type on keyboards where I could not switch the layout), so I decided it would be a career enhancing move to quit Dvorak and go back to QWERTY.
The pandemic situation has seen people heading in the opposite direction of travel from that you describe: from daily attendance at an office to predominantly working at home, if at all. The expectation in the near term is that working from home will remain much more prevalent and these people have the opportunity to set up home offices where compatibility with their co-workers is less of a factor. Will we be seeing a swell in the number of people choosing to train themselves on alternative layouts?

Post Reply

Return to “Keyboards”