Pros and Cons for different layouts

Typing & Ergonomics
21

Thank you, that is certainly interesting, however those improvements were measured by analysing English prose. If your everyday usage is different, especially if it’s a significant difference (programming, or a different language, which might not even be related to English and might make use of accented characters for example, that are by default not even present in these layouts), then the layout change might not give you a similar benefit.

In my case I think the costs (being less efficient on random keyboards in the world; angering my wife if she tries to type on my keyboard; having to research which layout is optimal to my current usage which could change; learning curve) outweigh the benefits I would expect.

2 Likes
22

FWIW, I’m typing a lot of Hungarian, with accents & all, on a Dvorak-based layout, and it is working like a charm. Not only do I type Hungarian faster than I did on HU-QWERTY, the accented symbols are at a much better place now, easier to reach, that put less load on my right pinky. I also do a lot of programming, same experience.

Point is, all of these alternate layouts are adaptable. You don’t have to use them as-is. You can rearrange them as you see fit, to make them more tailored to your needs. Just need to select a reasonable base. I took Dvorak, and moved the symbols I use often to places on the keyboard that are easy to reach, making my layout Dvorak-inspired in the end.

4 Likes
23

Thanks, that’s useful to know. I might have a look at your layout some day!

24

Imre,

Do you use a standard Hungarian keyboard?

25

I use Hungarian QWERTZ and UK English QWERTY and use the OS to switch between them.

26

You may be just the Hungarian language tester I’ve been waiting for. :slight_smile: DM me if you want to try a custom layout that might be optimised for your use case. I feel guilty about spamming random threads with my hobbyhorse.

4 Likes
27

I’ve used Dvorak, Workman, Colemak, and Norman. Each has their pros and cons, but I’ve stuck with Colemak.

Dvorak is, IMO, an “outdated” alternative. Better than QWERTY, but it’s very old and other, superior layouts are now available.

Workman is a good layout. It is a very good layout. What I didn’t like about it is that same-finger usage is a full point higher than with Colemak, and one very common digraph (“ly”) is typed using the same finger. While it’s better for “th” than Colemak is, in practice I found “ly” on Workman to be far more annoying than “th” on Colemak (which I don’t consider annoying at all).

Norman was … pretty good, but it’s designed for Norman, not for me. His rationale for the layout make several assumptions that simply don’t apply to me.

Colemak has two big advantages: fewer changed keys compared to Dvorak, and it preserves the ZXCV keys. A third advantage is that it’s included by default in Windows, macOS, and even iOS, which can’t be said for Workman or Norman.

This page has an interesting comparison of what it would be like to type different books on QWERTY, Dvorak, Colemak, and Workman. Workman is the winner in the “distance travelled” stat for each one, but it is the official Workman page, after all, and I don’t think it’s the end-all be-all stat. More important, I think, are the Home Row and Same Finger stats, which is where Colemak wins.

2 Likes
28

Colemak is not included by default in Windows 10. I just tried to install it.
Thanks for the rundown; I’ve chosen to try Colemak and found your points very reassuring ;).

1 Like
29

Dvorak is, IMO, an “outdated” alternative. Better than QWERTY, but it’s very old and other, superior layouts are now available.
Colemak has two big advantages: fewer changed keys compared to Dvorak, and it preserves the ZXCV keys.

When I was learning Dvorak in 2006, there were no other alternatives. If Colemak was available back then, I probably would have chosen Colemak over Dvorak, though not for the reasons listed.

It’s that the right hand does more work than the left one.

3 Likes
30

Huh, I could have sworn it was, at least at some point.

31

Not only that, but the data is edited, finished prose. It doesn’t measure actual typing, so no consideration is given to things like backspace, tab, enter, or cursor movement, which can figure very frequently in real use.

3 Likes
32

I’m not sure that makes much of a difference. AFAIK, none of the layouts discussed (save for one Colemak variant, which moves backspace) touches modifiers, etc.

1 Like
33

I am using a custom layout, and I think it may be interesting to you because you seem to be from Poland :wink: and I am, too. Here’s the alphanumeric part:
keyboard-layout
(Compared with QWERTY, keys marked in red use different hand, and keys marked in yellow are moved to a different place within the same hand.)

It’s based on Norman layout, with some tweaks:

  • K and J are very infrequent in English but considerably more frequent in Polish. Their position in Norman (middle of top row) is very uncomfortable for me to reach, because I have short fingers, so I moved them to better places (I’m particularly happy with K, because it landed just next to its original QWERTY location).
  • I kept O and I in their original QWERTY location - for me these locations are not that bad, and this made it easier to learn.

As for learning this layout, I don’t remember how long did it take me, but I remember one thing: relearning the 2 letters that changed hands caused me as much trouble as relearning all the other 11 letters that changed keys without changing hands. Now, in Dvorak 22 letters change hands…

3 Likes
34

Good point, except that this topic is not limited solely to those layouts, since we’re discussing custom layouts as well. Furthermore, when a typist makes corrections, it’s not just the backspaces that are lost in the text, but anything else typed, as well. That could lead to significant differences in the frequency of certain letters. For example, t probably gets typed and erased more often than most other letters.

I’m interested in the actual frequency (and pattern) of characters typed, which is a much trickier problem than analyzing finished text samples; it requires a real keylogger, and even that isn’t quite right, because the keyboard could be sending multiple events with a single keystroke, and if the goal is to make a better keymap, both the input stream that the host sees and the keys pressed on the keyboard are interesting (and not necessarily identical).

3 Likes
35

Indeed, that’s an important point.
I started playing with keyboard layouts a long time ago (10 years?) but didn’t get serious until I got RSI. And, from what I can tell, it was moving the modifiers that made most difference for me in terms of reducing discomfort. So, let me update the diagram with what I did to modifiers:

keyboard-layout (1)
keyboard-layout (1)830×290 30.9 KB

As you can see, Shifts are closer (moving them upwards really makes a difference, especially in case of right shift), Ctrl is under thumb. I couldn’t find a good spot for backspace, because there is not enough thumb keys on a typical keyboard, so I put it under B - this required me to move B and make some changes in punctuation (I access most of it using right alt). This part is not optimal and subject to change when I get my keyboardio.

Going back to the benefits: for me, changing alphanumerics was mostly about making typing less difficult. It probably improved my speed somehow, but what’s more important is that my fingers can move more fluidly, and it’s just more pleasant.

I suppose this depends a lot on finger length. People with long fingers can easily reach all letters so for them it’s not much of a problem. My index finger is 66 mm long, and middle 77 mm long (even though I’m 6 feet tall), and reaching QWERTY’s T (and especially Y) is a big stretch - my whole hand has to change shape (flatten) to reach that key, making all other fingers move from their positions. That’s why I put punctuation there.

6 Likes
36

This is a very interesting approach, and I’ll be very curious to see how it develops. It lines up with my observations as well: I think the best way to judge a layout is to assign scores to ngrams instead of just keys. I guess 3-grams should be enough for good results.

3 Likes
37

Indeed, the carpalx tool which Jesse linked above, analyzes typing effort in terms of 3-grams. However, it uses much different criteria than AmigoNico does. (AmigoNico - excellent writeup, by the way.)

2 Likes
38

I am pretty expressly planning to do some remappings of the non-QWERTY keys, that is, i intend to retain the ‘official’ portions of QWERTY, but i anticipate attuning the others (e.g. - ctrl, bksp, cmd, space, alt, and so forth) to my own intentions.

Enter is already swapped onto the butterfly key; I use it plenty enough that it needs to be that bit more direct.

i have a planck mapping where i shifted almost everything but the “official QWERTY bits” hither and thither. i’d suggest looking at it except that the Planck is sufficiently smaller that it is guaranteed that it seems unlikely to provide any wisdom.

i’m starting by remapping the things that are bugging me… So far…

  • Enter’s gotta improve…
  • Space isn’t right…
  • i think I need shift to be elsewhere
  • NUM controlled at top right? Nope, nope, nope… maybe backspace instead…
  • Left pinky shouldn’t be pageup/down…

The one unfortunate thing is that with keys being pretty unique in shape, they can’t be readily swapped as I could do with “regular” caps. in the long run, i suppose i want runic caps :wink:

2 Likes
39

i have tried dvorak, colemak, and now norman.

in between i’ve always gone back to qwerty then to the new layout.

i didn’t like dvorak for the alternating keys. once i got really comfortable it seemed like i was switching common pairs that i would type really fast.

i like colemak but it was starting to get where i was really fast with it and then i didn’t like it anymore. i guess i don’t know why i switched away.

norman is actually my least favorite. i think that it’s so close to qwerty that i make a lot more mistakes. i cannot explain it but i am thinking about switching away to maybe workman or to colemak-dh.

1 Like
40

I had already done a bit of research about proper posture and hand placement and followed those guidelines for a few years prior to switching. I only switched to colemak because I could think of nothing else that would allow me to keep using a keyboard without experiencing pain.

Colemak was great for a few years, the old pain went away but after a while I started feeling discomfort on my right hand’s index finger. After doing some more research on colemak I figured it was the placement of the letter H. I then switched to workman.

Workman has been working great so far for the past couple of years, the only issue I ever had with it was when I needed to work on a terminal outside of X on linux and it wasn’t that easy to get it working as with other more popular layouts.

To me it seems minimizing lateral movement to be the most important thing about a layout. I know both colemak and workman aim to do this to some extent but all their testing was done with the old key placement of most keyboards. But now with the M01 I’m wondering if the difference in having straight rows for each finger has an impact on the formulas that were used to come with the optimal layout for english with the least amount of lateral movement.

7 Likes