Hey lolop, thanks for entering that into the analyzer. That's really helpful and interesting.

If it has to have a name, I prefer something that includes the word Colemak... perhaps "inColemak" as an abbreviation for in(cremental) Colemak. Really, it is a learning process and not a competing layout.

That said, if DreymaR wants to promote it as being Tarmak-related, I'm fine with that. It is a variation on his idea. It could be called "Tarmak4p" (Tarmak for programmers).

Whatever gets the idea out there is fine with me. I just hope that someone finds it useful.

Hey DreymaR, I edited the original entry as you suggested.

I analyzed some code and ';' comes up between 1-2%, which makes it as popular as many of the other letters. Personally, I think that I would find it very disturbing to have it in the middle of the keyboard.

I started with EKN and GTF due to their large benefits (as you described in your post).

I also felt that it was very important to finish the intermediate stages as quickly as possible. At the beginning stages, the benefit of having letters in ideal locations is less significant that the challenge of the learning process itself. While it might seem good to optimize according to Codesharp, it has to be weighed against the frustration of reinforcing the wrong letter location. At each intermediate stage, 25% of the moved letters are in the wrong location. This is why I left the self-contained LUI loop to the very end. In fact, I recommend that you consider that for Tarmak. The additional benefit is that it will disturb the programmers less because the ';' will make it home sooner.

I understand that it can be confusing to have too many options. I just wanted to put the idea out there for consideration. I don't feel a strong sense of ownership over it or that it has to have a particular name. If, by chance, any of it is useful to you, feel free to take what you like. I appreciate the work that you've done on Tarmak and I do not plan to put any more work into it.

This is a tangent, but do you think that there might be a better chance of getting Tarmak on the front page if it had a name that sounded more Colemaky? My first impression, from hearing the name "Tarmak", is that it is a competing layout, which is really not the case. Again, just putting the idea out there.

Cheers!

I see that the latest news on the front page is from 2009. Obviously, the front page is dead.

Lalop: Does Karl's step 3 have an edge over Tarmak3 in Patorjk's analyzer? Do you have a link to your Tarmak analysis too please?

I found it very amusing to watch the heat maps move towards the home row with each step! I think we should make an animated gif out of that. Again, if I could see an analysis of Tarmak analogous to the one you made of karl's steps please?

Okay, I did a side-by-side comparison on Patorjk's analyzer of a bigger snippet of text: A couple of chapters from Pride and Prejudice (found at Project Gutenberg) containing more than ten thousand strokes of the letter E. Should be good enough.

Interestingly, the 141 instances of J prove that it is indeed a rare letter and the rarest of those that are moved by Colemak. Even the colon/semicolon key (241) and Z (142) had more presses and only X (102) and Q (83) of all the letters less presses. Another good argument for sticking to moving the J and leaving the semicolon out of this?

Tarmak(ETOI): http://patorjk.com/keyboard-layout-anal … d/2b1Bz5t9
Tarmak(ETRO): http://patorjk.com/keyboard-layout-anal … d/MVB2CBnK

As in my Tarmak post, I use the ETOI vs ETSO/ETRO nomenclature to show which important letters are brought to the home row at each step. Makes sense to me, but maybe not to everyone. ;)

Again, karl's steps win this analysis both on step 3 and 4! I'm getting more and more convinced here. Also, it seems that the ETRO scheme evens out the benefits per step better.

Patorjk favorises hand alternation which brings Colemak to somewhat of a disadvantage per se since it focuses on same-hand rolls instead. I don't know whether this bias will play into the present analysis but I shouldn't think it matters a lot? The Tarmak(ETOI) scheme moves keys predominantly on one hand at a time (R-L-RL-R-L), and so does the Tarmak(ETRO) scheme (R-L-L-RL-R).

[edit: Oh lookie, heat map animations below!]

   
   Animation of Tarmak(ETOI) step-by-step heat maps from Patorjk's analyzer based on a largish English text sample.

   
   Animation of Tarmak(ETRO) step-by-step heat maps from Patorjk's analyzer based on a largish English text sample.

Yes. Whether it's simple line endings or sequence delimiters, both colons and semicolons get their share of the action in coding. I was just fascinated to see that they're still 70% more common than J in an English literature sample. It does provide an additional argument for his way of sorting the Tarmak steps, even for non-coders but all the more so for those of us who do type in codey stuff!

Yes, I agree, if I were using the wide mod, then I don't think that I would find it jarring to have the ';' in the middle.

While I tend towards more generic names, I think that the Tarmak name hasn't hurt you. It's probably helped. Even in writing these forum posts, it's helpful to have a handle to refer to. I also find that it sticks in my head.  ;)

About the name of this 'J' hopping sequence:

If, in your detailed analysis, you determine that the above set of steps is more optimal, I'm totally fine with it being absorbed into Tarmak, either as the default or as a programmer-specific choice. I know that it's complicated and that, at this stage of the game, every change means losing something in another area. Besides, I didn't present a dramatically different idea. I just refactored the big loop to eliminate the ';' hopping and moved LUI to the end.

If, on the other hand, it ends up standing on its own and you refer to this set of forum posts as the main reference, then I think that we should stick with the name "Incremental Colemak".

Further comment about the frequency of 'J':

One weakness of this approach is if someone, for whatever reason, does have the reason to type a lot of J's. For example, what if a guy named James Jordon Jr. writes for a jube jube factory in Jumbalaya, IL? I still think that Incremental Colemak is better for programmers, but I can imagine that some small percentage of the population could have a problem with it.

Tangent...

If it were up to me, I would be in favour of removing S from the loop and leaving it permanently at the QWERTY position. I know that it's less optimal, but it would be easier to learn and live with in a QWERTY world. That said, I believe the strength in sticking to the standard is a greater benefit for most people.

[edit: added jube jubes. oops, S not R]

On the contrary, LUI is a huge gain (it goes from being "worse" than your step 3 to "better" than your step 4) so long as it takes place after the O is moved.  It also costs only three letters.

Er, AFAIK, only the semicolon, a rare letter, is potentially "burned in" by having LUI as step 4.  If you are talking about letters differing from colemak in general, then you should consider that LUI (14.37%) are typed almost as much RSDP (18.75%).  Being able to achieve the former much more easily (for the cost of 3 vs 5 letters) is a significant advantage.

It's a significant advantage for everyone who doesn't.  If they do, probably your scheme is a far cleaner solution.

Your proposed default makes life slightly easier for the C-styple programmers (remember, they'd still incur the disadvantages of doing that 5-key stage first) at the significant expense of everyone else.  Sometimes, it's better just to accept that something's not designed for your usecase, and go with the other thing that is.

That reminds me, will we be seeing an implementation for whichever platform you're using?

As mentioned, I don't think that the semicolon would get "burnt in" the same way that a letter does – even a rare letter. To me at least, the symbol keys work a little differently. Maybe it's because they're usually not hit in polygram sequences but alone. This fact on its own would suggest moving the semicolon around all the time instead of the J, but there are other considerations. That stuff about potential J users doesn't worry me though. It's quite hypothetical and if such persons exist they'll just have to make their own variants.

Tarmak's full name is "Tarmak(ETOI) transitional Colemak" or a variant of that, which shold be descriptive enough. "Transitional" covers about the same meaning as "incremental" in this case I feel, and it's the name Shai suggested. I'll refer to your scheme as "Tarmak(ETRO)" but for the user that's too specific I guess so some confusion might arise. Also, since the I loop is in focus now maybe it'd be better to use the full progression and call them "Tarmak(ETOIR)" and "Tarmak(ETROI)", or drop the I and use "Tarmak(ETOR)" vs "Tarmak(ETRO)"? Hmmm... :)

Tarmak has already been through one upheaval: I started out recommending the LUI loop first(!) since that's easy and quite beneficial, doesn't mess up any key positions and I thought it'd be a nice warmup for the big loop. Then I did a right-hand loop moving only the semicolon over to the QWERTY E position and finally a left-hand loop. That has an aesthetic appeal and simplicity to it, but some analysis showed how much there was to gain from moving GTF. Also, making a few more steps allowed the user to select an even more incremental scheme while the braver souls could still skip a step or two (I think I'd start out with step 2 myself, to get a little challenge for each hand and a lot of benefit for it – after that it's more of a toss-up for me).

As mentioned, I think I'll make an implementation that lets you easily choose when to do the LUI loop – although one setting has to be the default one. I know how to do it for Linux XKB and the Arduino solution from GeekHack (which I haven't used but love the idea of!), but for Windows PKL it's a bit more hassle and confusion as always because it hasn't got "layers" to work with. Not a biggie though.

Looks to me like a choice between ETO(I)R and ETRO(I) schemes with a flexible I placement is a nice possibility. Not sure what to recommend as a default though.

Anyway, I've made a new figure to show the ETROI steps in comparison with the old ETOIR ones:

   
   Fig.: The Tarmak(ETOI) transitional Colemak layouts, spectrally color-coded from red (Tarmak#1) to violet (Colemak).

   
   Fig.: The Tarmak(ETRO) transitional Colemak layouts, spectrally color-coded from red (Tarmak#1) to violet (Colemak).

Incidentally, the IE/EI same-finger bigrams are a problem with the Tarmak layouts in general. This is an argument for an earlier LUI loop but as lalop points out it should likely be done after the JYO partial loop so that you don't replace the IE/EI problem with a IO/OI problem! The individual learner should probably consider how long they plan to be sticking with each step, as these same-finger bigrams may be tolerable during a learning process but detrimental over time.

[Edit: I've had a naming crisis in which I went back to referring to the RSD step by 'R' rather than 'S'. This is because R is brought to the home row but S is shuffled to a maybe slightly worse position so the focus shouldn't be on that letter. Sorry for the confusion.]