Test: QGMLWB faster than Dvorak?

[qgmlwb]

This intrigued me: Mullenweg is a Dvorak Keyboard user and can type over 120wpm   

I have to ask myself why am I using QWERTY? Keyboard is the device that I use to talk to the computer, is QWERTY slowing me down?

So I decided to learn Dvorak, I do my research apparently it's worth it:
http://blog.hanschen.org/2010/01/30/dvorak-two-years-later-was-it-worth-it/

I try to learn more from others before I dive in:
http://arstechnica.com/gadgets/2014/03/my-quest-to-learn-the-dvorak-keyboard-layout-part-1/ From her negative experience using it without practice I decide to learn it more systematic ways:
a) practice the letter groups. When learning qwerty we started with asdf combinations only! Then jkl; son on ...
b) have all letters in my muscle memory before trying to use it in real life - she kept forgetting letter locations which frustrated her.
c) she found some fault with Dvorak in part 3 so I searched to find if there are any better layouts out there. I found this: http://mkweb.bcgsc.ca/carpalx/?full_optimization

So Dvorak is out, I am going to learn QGMLWB

So I challenge myself that in the next 10 days I'll practice a hour a day and learn QGMLWB so that I type at least 50 wpm (where I consider a word as 4.5 characters+space).

Day 1 goal: learn all 30 character locations 

stay tuned ...

[mstechfreak]

If you want to learn DVORAK, I'd suggest you to use TypingStudy  - http://www.typingstudy.com/en-us_dvorak-3/lesson/1

They have all kids of keyboard layouts there. I did it myself and it was really useful. At least you can try, if you don't like it, you can always switch back to QWERTY.

[Oobly]

If you do some research, you'll find there are many, many custom layouts out there (Simplified Dvorak, Programmer's Dvorak, Colemak, Workman, Neo, AdNW, QGMLWB, etc, etc). I suggest you read up on people's experiences with using various layouts before deciding, too. For instance: https://geekhack.org/index.php?topic=74943.0

The problem with using a layout that has similarities to QWERTY and using the same physical layout is that you're essentially retraining your QWERTY muscle memory to the new layout which can really mess you up when switching between them (as many people will need to be able to do for various reasons, such as if you're visiting someone and need to use their PC, etc.).

Which is why I recommend going for a fully customised, fully optimised layout, tailored to your own preferences (alternation vs rolls, etc), but only on a different physical layout board. The physical layout has more benefits in terms of efficiency, ergonomics, RSI, etc, than the character layout.

I recommend the AdNW software for analysing and customising your layouts as you can alter the weighting of different optimisation factors (key position distance factors, inward rolls, outward rolls, alternation, same finger, etc) and provide your own corpus for the analysis and optimisations.

Keeping ZXCV positions for common shortcuts is not a good thing to do, either. It retains too much similarity to QWERTY and can really mess with the optimisation of the layout. I have found it very easy to learn the same shortcuts on my own layout board as part of the training process.

That way, you keep your QWERTY typing skills in place while learning (and normal QWERTY typing is still very useful to be able to do for a lot of reasons) and don't suffer if you have to switch between them. Eventually you may lose a little proficiency with QWERTY due to not using it for long stretches if you're using your custom board most of the time, but it's not something that'll ever fade completely.

You can do some basic comparisons of different layouts here: http://patorjk.com/keyboard-layout-analyzer/#/results

You can post your own text into the main field and run the test. There are LOTS of existing layouts for you to compare there (including QGMLWY, but you can easily modify an existing layout or make your own for testing, too) and the statistics after running the test can be quite revealing, particularly the finger / hand balance and heatmaps.

[qgmlwb]

Thank you Oobly for your reply and suggestion, I appreciate it

I agree with you on the physical keyboard change, but my worry is that I buy one and then I will be too lazy to practice a new layout on it. So if I reach my goal of 50 wpm then as a reward I am thinking of buying Typematrix EZ Reach 2030

For changing the keys I have used The Microsoft Keyboard Layout Creator (https://msdn.microsoft.com/en-us/goglobal/bb964665.aspx). tip: you can lick on the key and change it, export it and install it as an msi.

Thank you for the patorjk link, I have tried to use it before but I wasn't aware that you can customize it. I've submitted the QGMLWB layout but it wont be approved it seems. Based on it's comparison using all 3 datasets Colemak wins with 205.48 points QGMLWB is 6% worse at 193.9. I've pasted some source code kafka-net concatenated and now QGMLWB won 50.60 vs 50.47 Colemak. It seems that the ranking depends heavily on the criteria they use and the dataset. So now I have a new question: which criteria or dataset to use?

It seems Colemak is the best standard layout, and I have considered switching to that instead. But I wanted to do an educated decision so I have reviewed and studied carpalx criteria and I have to admit it makes a lot of sense. I have read at a different blog that at high speeds same finger repetition is the biggest blocker so I was particularly paying attention to this metric (stroke path). QGMLWB seems to beat Colemak in that race. I would have loved to review patorjk algorithm as well but unfortunately is not open source so I cannot see how sophisticated it is.

So based on my review I will stick to QGMLWB for now until someone convinces me otherwise. I have wrote in a few minutes a very basic (windows) training app that allows me to practice in QGMLWB on a qwerty keyboard by typing a random key displayed on the screen. You can enter the keys you want to practice which was very useful to learn keys in chunks. It cycles through all letters randomly with the same probability so this allows me to learn all keys evenly.

Result for Day1: now I remember all keys on the new layout

Goal for Day2: type words instead of letters

[qgmlwb]

Day 2: Letters are getting boring and seeing only one letter at a time slows me down. So I change the app to do 2 letter combinations. This is much more fun, it seems I have grown out of the 1 letter mode, we have to keep up the challenge with the skill level. So if 2 is better can we try more? Since my target level is 4.5 I bump it up to 5. I like this, typing is much more fluid. Some letters come easy some letters I have to search my memory where it is in a few seconds

Progress is slow, I can't get the speed up, it's not that I don't know some of the letters, but I have to stop and remember most of the letters, it's not in my muscle memory. OK I try to do only vowels, 5 letters, how how hard can it be to move it to my muscle memory? Hard, can't do it. OK I reduce it to 3 letters A,E,O they are on the home row next to each other, can I move them into my muscle memory at high speeds easily? No. It seems there is no substitute for just getting though it with sheer practice. I go back using all letters  )

I try to push myself to type faster than I can, this helps, I make more mistakes which is OK as long as it helps with the speed. There is another problem, I keep trying to correct the mistakes, now this slows me down. The word turns red if I made a mistake and green if it was correct, I can't stop myself focusing on correcting the typos even if I know that my penalty system will mean that the wrong letter will be repeated in the next 1,4,12,30 characters. I am thinking of adding a time penalty so that it speeds me up.

I want to know how fast I am typing first so I add counting the words per minute. I get up to 8 words. This is my typing speed after 2 days, not bad, but I think this was the easy bit, it seems it's going to be harder to improve beyond this point. Seeing my WPM also motivates me to stop correcting my mistakes, great!

I have to improve 625% to reach my goal. I only practice for max a hour a day. I have to improve every hour 26% to get there! Is it possible?

Will let you know ...

[qgmlwb]

Day3, had to skip a few days of practice There is only one ting that bothers me about QGMLWB all the vowels are on the home row on the right hand key except the U, this makes it hard to remember and I have to make a mental exception when I am typing U. carpalx punishes the little finger too much. I'm glad Casey has criticized Dvorak on her blog, this gives me the power to criticise carpalx and the courage to swap H and U. I can do this before I begin my 3rd day of practice I hope. Is it too late?

First thing I have to measure if the swap wouldn't make this keyboard a lot worse than the semi optimal found by carpalx. I don't have time to run a simulation by carpalx so I just fire up http://patorjk.com/keyboard-layout-analyzer/ suggested by Oobly, thanks for the tip again. I set up the keyboardand and it seems that the new layout is a tiny bit better than the old one, hurrah! It's a go then I will hack the layout and create my own  This makes me feel a bit geeky, I am going to learn a layout that nobody has ever tried before  (at least to my knowledge), this is an uncharted territotry!   I call it the QGMLWBH layout. It halso makes this post more worthy to post on geekhack.


I reduce the training set to IAUEOH and focus on that only. OK I have to admit that I did make a few mistakes of mixing U and H but I like this layout better, it feels much more fluid. 

I go back to the full 30 key set, see how far I can push it today. It feels like I'm hitting a wall, there is a a speed that I can't get easily beyond. It is far from being automatic, I still have to put a lot of effort in remembering the keys, I have to do it all consciously my unconscious mind is not helping. At the end of the day I reach 10 WPM, a little disappointing, but this is what I've got. On the positive note, I am not - yet, noticing any mistakes on the QWERTY layout from my little experiment... 

[Keymonger]

Pro-tip: focus on ease of use, not speed. Speed sounds attractive when you hear about high WPM and such but ultimately comfort matters most, and that can't be expressed with a number like WPM. I'm pretty sure if you hold your hands over the keyboard in a very tense manner you will type faster than if you were to type with somewhat relaxed hands, but with tense hands it's not very comfortable. Typing very fast is like sprinting. I don't know what you type and how you type, but in my experience typing is very rarely like sprinting, and oftentimes more like a brisk walk.

It fascinates me that the main alternative to Qwerty seems to be Dvorak still, but I personally think it's an unacceptable layout just for the position of the L key alone. If you touch type, your pinky is supposed to hit L. No thanks! In English, L is used far too often to justify that. Colemak seems better, but there's a few things I don't like. A misguided adherence to Qwerty for compatibility, which ultimately is a kind of pointless semi-compatibility. And maybe a bit too much emphasis on alternate finger usage, which in theory sounds good but IMO there's more to a comfortable layout than that.
I remember when Workman was introduced and the reasoning behind the layout was elaborated upon by its creator, it sounded good but on that layout the T is to the right of H, which doesn't allow the fingers to roll inward when typing 'th' which is very common in English. I remember a few years ago I wasn't happy with Workman either but I was inspired to design my own layout with the above analyzing tool. It was pretty awesome. I seem to have lost it though. Oh well.

Anyways my suggestion would be to focus on comfort. I don't think I would ever recommend Dvorak over Colemak or Workman to anyone.

[MajorMajor]

IMO the main benefit of using DVORAK over Colemak or any other alternative to QWERTY is the ease of use - you can go to any computer with a modern OS and find DVORAK ready to go. No need to download any files or remap keys. Colemak hasn't been shown to be a big enough jump in efficiency to warrant giving up that privilege, IMO.

As for the right pinky use, I think if you're willing to learn an entire new keyboard layout, you might as well learn how to type with all of your fingers. I don't find any discomfort in moving my pinky up to hit the L key, it's very natural.

As for OP's goal

So I challenge myself that in the next 10 days I'll practice a hour a day and learn QGMLWB so that I type at least 50 wpm (where I consider a word as 4.5 characters+space).

I gotta say, there's no way you're going to get to 50 WPM in 10 days even if you used the layout 8 hours a day, let alone 1 hour.

It took me 6 weeks of using Dvorak 8+ hours a day to get back to 50 WPM, and I previously typed ~90 with Qwerty. The muscle memory is very hard to break down and rebuild! Now 7 months later I'm back up to around 85-90 WPM.

It's a terrible and slow first 2 months, but I think you need to jump in full-time in order to make any serious progress since so much of it is mental memory. You're not going to progress quickly if you're still relying on Qwerty most the time.

[Zustiur]

I taught myself how to type colmak in December. I used a typing website that I can't link you to right now. It just had words of certain finger combinations, no fancy measures or accuracy checks.

I tried to only use colmak so that I could learn as quickly as possible but the loss of speed was too frustrating and I gave up completely for a couple of weeks. When I started again the memory of key locations was already in place and I retained most of what I'd learned.

My next practice method was to take a long document and copy it out. This seemed to work well as a learning aid because it was a document that I wanted a copy of anyway so the goal was reinforced because I was achieving two things at once. This got me to about
25 wpm.

From there it was a long slow climb for speed and accuracy. 6 months later I finally had a good enough speed to start using colmak at work. Now I'm at roughly the same speed on both because my qwerty speed had reduced slightly. At a guess I'm at 50-60 wpm but I haven't measured it. That's about as far as is practical for my job anyway. The really high speeds are only useful if you take dictation or do a lot of copy editing.

My advice is to not set your goal of X wpm by Y hours of practice. That will only dishearten you if you don't succeed. To make goal estimates like that you have to have a lot of experience of doing similar tasks, which you don't have here.

I highly recommend copying out a book or document that you want in electronic form. The benefit of having a goal, other than just typing itself, is huge.

[Zustiur]

Oh, and some advice from my typing teacher many years ago; focus on accuracy not speed. Every mistake you make costs a lot of time. Fixing one word could cost you two or three more in speed.

[MajorMajor]

One tip I have - use a WPM tracking website to measure your progress, it will encourage you to keep improving. I used 10fastfingers.com, just their basic speed test.