The short answer is no, the whole point of Mod-DH is to be close to vanilla Colemak, with only minor tweaks.

Hi Nickel,
I'm guessing it's just the way MS Keyboard Layout Creator writes its output file. Obviously it doesn't matter to the program what the keyboard actually looks like, only what scan codes come in from the hardware. For those extra, non-letter keys, the main thing is that the OEM codes should be mapped correctly. Let me know if they aren't. There may be some regional variations for those too, so of course you're free to edit them to your preference.

In MSKLC you can set the keyboard type to conform with your actual keyboard. This is independent of layout files.

I found M was the trickiest change to adapt to. It was the only key that ends up in a significantly worse position, which is probably why. I think it's quicker to learn the new position of key when it moves to an *improved* position. The new H, for example, was much faster to learn.  For me, the K stayed in its standard Colemak position of course, but I do still find LK bigram slightly annoying, and I think your K position would improve it. That said, I guess my LM bigram is better, so swings and roundabouts.

The B was a little tricky, changing from bottom to top row seems like a big jump... But I really disliked the old B, coming from a previously non-angle-modded configuration, so I think it was much easier overall. Worse than the B for me, was actually the C. Since I applied the angle mod for Z,X,C at the same time as I did the Mod DH switch, C suddenly moved to a more awkward key.  It may also be that I adapted more easily since I had been typing standard Colemak a shorter period of time than yourself.

I think the experience of switching to a DH mod would depend significantly on (i) whether the user is already using the angle mod and (ii) if they are not, which finger they are using for C.

Yes I did use alternative fingering for C - I used my index finger for years without really noticing it. Hence my initial reluctance to apply the angle mod.

Probably for a more frequent key, the amount of elapsed time it takes to learn the new position is likely quicker, but that's not the whole story... you'd need to think about how many errors you made in that time period as well.

The measure of the impact of changing a key should be something like:
(total number of errors made caused by the change) x (average time taken to correct such an error)

For example, we talked about moving the J before. It no doubt takes a longer time to relearn J because one practises it so little... but the question is, how much typing-error time does this lead to? Not very much I guess. Changing E would cause you to make a lot errors initially and you'd learn it faster, but would the total number of errors made as a result be fewer or greater than for J?

Just reporting back after more than a year with my variant of the angle mod-dh as discussed earlier in this thread. It was great while it lasted. What changed?

Building a Planck 40% ortholinear (matrix) keyboard! I will not use anything else now (famous last words)! The perfect symmetry and shorter finger travel to all adjacent keys makes even the angle mod feel like a compromise, to me. And incredibly, the 40% layout makes even the 60% Poker 2 keyboard feel cumbersome and unwieldly -- which is achieved by the ability to flash more layers than one can possibly use.

I introduced only one small change to the mod-dh (and, colemak, in general), and that was to replace the slash key with the quote as can be seen here.

It may look a bit weird. Where are the shift keys! One of the cool features of customizing the keyboard firmware is the ability to impart keystroke behaviours. In this case, modifier keys defined in my outer keys (bottom and sides) are enabled if held down (versus tapped) as can be seen here. The thumbs by default handle space, tab, backspace, delete, left and right shift, and activating the symbol/number layer.

All sorts of behaviors like double tapping, single shot functions, etc. can be introduced. A more verbose writeup of my particular Planck configuration can be found here

P.S. Building these little suckers with various key switches and caps is fun too!

Hello Steve. The angled DH mod which felt very good on my previous staggered ANSI keyboard, feels even better on a matrix keyboard. The grid layout makes the keys feel much closer together, so the DH feel very comfortable indeed. Even the lateral movements like the M feel less of a stretch.

Adapting to single width modifiers took very little time to get used to. Because on my previous ANSI keyboard layout I was already using my left thumb for space and right thumb for backspace (using the right Alt), I was already "thumb" biased, so to speak. So to be able to add extra thumb actions because the usual wide space bar is absent was an exciting prospect for me to test out. It works for me. I had no difficulties developing the finger memory for the thumb actions I assigned. If you look closely at the image of the Planck I posted, you will notice I also rotate those thumb keys so they are angled downwards for a more natural contact. The Planck can also be built with a 2u wide keycap for the space key but I wanted an extra thumb key to preserve symmetry for my right thumb actions. In actuality, because of the perfect symmetry of a matrix layout, and the minimal travel required of the fingers and thumbs, I found it very easy to adapt to. So much so, I find it difficult to use a standard staggered keyboard which now feels very unnatural.

It did at first, feel a bit tight. Now, on my previous ANSI layout, I never did the wide angle mod, just a right shifted angle mod, so my hands in actuality were essentially the same width apart, so I was used to that. It was more the visual effect than anything else, the Planck takes up so little real estate. The comfort of the matrix layout compensates for the closeness of the hands immensely --- especially compared to my previous ANSI keyboard. BUT I am looking forward to building a "split" Planck, two 4x6 keyboards, which I am told is in development. Still, there is something to be had with such an effective diminutive package you can just grab and plug into another computer that has whatever layout you want immediately available.

On a side note, I programmed a Workman layer on my Planck to try it out because it supposedly was optimized for matrix style keyboards. In all honesty I didn't stiick with it very long, but I much prefer your DH mod on a matrix layout. The increased same key fingering of the Workman felt very unnatural to me.

That's a nice suggestion and a nice initiative!

One problem is that there are several options. There are two DH-mod variants and up to four+ ways of implementing the Angle shift. So it can get a bit cumbersome.

I like every swap from colemak-dh really. I've started using an angle mod which was the other massive improvement. I think the work both of you have done with these curl mods is great.

Well I never used vim or related software with qwerty, but I've always liked using the qwerty hjkl positions for directional commands. I haven't been actively following discusisons here lately, so I may be misunderstanding what you mean. Do you mean that you think the locations of the directions in extend are superior? If that's the case, then I'd have to disagree. For my use of directional actions, the qwerty hjkl position are more optimal. They place the two more used directions (up/down) in better positions and not on the same finger. Or do you mean something else?

I always boot straight to console, so it matters a little more to me. I can still type my password fine with vanilla colemak, but in the rare circumstances when I need to use the console for something else, vanilla colemak is hard to deal with. I'll probably make a pull request soon then.

I imagine it would be simpler to generate all those from the main keymap files too. I'm not sure I know all the variations though. I only know the 3 ansi ones listed here.

I could definitely add console keymaps for them in the current pull request or a new one. I can add wide versions of the ansi angle mods (excluding awing) for both too. I don't know if it's a good idea to add a keymap for every different possibility for what goes in the old qwerty b position though as it would add a lot more keymaps and it is easy enough to change this on key. Maybe I could add an option to manually generate such keymaps with user-specified changes.

Two and a half years in now! And I think, time to reassess? Thinking time, again. ;-)

I'm unhappy that we have two competing DH-mods. Makes implementations a hassle and confuses newcomers. But of course, there are multiple preferences and considerations and it's hard to strike the compromises that make everybody happy enough.

 q  w  F  P  B  J  L  U  Y  ; 
  a  R  S  T  g  K  N  E  I  O     The Colemak-Curl(DbgHk) layout – my current Curl-DH mod
 z  x  c  v  D  _  H  m 

 q  w  F  P  B  J  L  U  Y  ; 
  a  R  S  T  g  M  N  E  I  O     The Colemak-Curl(DvbgHm) layout – SteveP's current DH-mod
 z  x  c  D  v  _  K  H 

Could we do it? See this post.

My proposal is this "best of both worlds, accepting just a little pain" solution:

 q  w  F  P  B  J  L  U  Y  ; 
  a  R  S  T  g  K  N  E  I  O     The Colemak-Curl(DvbgHmk) layout
 z  x  c  D  v  _  M  H 

 q  w  F  P  B  J  L  U  Y  ; 
  a  R  S  T  g  K  N  E  I  O     The Colemak-Curl(DvbgHmk) ANSI-Z layout
    x  c  D  v  Z  M  H 

It's influenced by the insight that to newcomers the Angle mod doesn't necessarily feel as an automatic single change. In the topic mentioned above, people chimed in that keeping the V in its old QWERTY/Colemak place seems at least as simple as moving ZXCV together. I've thought of it differently because I did the Angle shift first.

In fact, using the above Curl-DH mod with the AngleWide mod as I'd do (I call it Colemak-CAW), both V and M would remain in their old unmodded positions! That's kind of nice.

The other major point is that the M in my opinion deserves a better position than the "dethroned" middle trench. The above position will be better for it, at least on staggered boards which are by far the most common ones (actually, I don't like pure matrix boards). This means moving one more key compared to my current Curl-DH mod, but you know – so be it. I've come to think that the benefits will be worth this price in the long run.

For the record, Wikipedia (citing R. Lewand) and the Cornell page of letter frequencies in English report these numbers for the letters involved:
D: 4.3%
G: 2.0%
B: 1.5%
V: 1.0% (1.0–1.1%)

H: 6.0% (5.9–6.1%)
M: 2.5% (2.4–2.6%)
K: 0.7% (0.7–0.8%)

These frequencies, of course, tell nothing about the efficiency of n-grams. We've all noticed how much nicer the important HE bigram becomes with both the current DH-mods.

Heh, judging from pure letter frequencies the V might have to move to the upper row! But that'd bork the Ctrl+V shortcut way too much. So no. Just no. Right...?

I've come to realize that what the left-hand issue amounts to is not as much two different DH-mods as two different Angle mods! I may end up providing one Angle mod moving V and one not moving it, for ISO boards. For ANSI boards it wouldn't make much sense to insert Z between C and V per se, but for ISO boards inserting B there would make sense from a pure letter frequency point of view. Doing it that way would add a modularity that'd let people choose freely between the current left-hand DH mods without having more than one main DH mod. There are already several Angle mods and that's the way it has to be I think, so the hit there wouldn't be too bad. I have to think about that one!

Benefits, comparing the above proposal to the current DbgHk (mine) and DvbgHm (SteveP's) mods:
+ One common DH-mod would be much easier to promote and implement! Curlers, unite! :-þ
+ Apparently, slightly better D and H positions than DbgHk's – according to many users' preferences
+ Better M position than DvbgHm's position, at least on normal-staggered boards
+ Even B is actually somewhat more common than V in English, so V shouldn't have a prime position
+ [Edit 2017-09:] V and M (and K as well) are kept on their native rows, and with the Wide mod even in their native positions!

Pains:
- Three keys are moved on the right hand, two more than standard Colemak, one more than DbgHk
- The XCV Cut/Copy/Paste block is split up, unlike DbgHk
- This makes Extend a little uglier, wedging the left mouse button between Copy and Paste

Ideally, you want just one colemak, not just one dh mod

If it is genuinely better then it needs to percolate upwards

But I'm not sure how the no.s stack up?

i.e. how many colemak users are there, and how many dh users are there?

I would imagine there can't be too many dh-ers - only a small subset of the no. of active forum users..

do you think i would be being pessimistic if I said maybe a dozen?

prob I have is that on an ISO with wide-mod there is no discernible advantage for either v or d keys, so breaking up zxcv would be a detrimental move

however, if you did amalgamate and effectively forced the move - I think I could pretty easily accomodate it

but as I said originally, the real question here is not amalgamating two dh variants, its amalgamating it all with colemak to give an improved colemak

I genuinely think it is an improvement, having used both original and dh colemak extensively now..