So, if I understand you correctly, you've changed the first step from EKN(J) to TFE(K) and then the rest of the steps are as the current Tarmak? A simple but interesting transposition of the first two subloops.
That'd make a Tarmak(EtNROI) progression – in which case today's progression might better be named Tarmak(EnTROI). (Damn, these namings...! Haha.)
Tarmak(EtNROI) said:1) The T>F>E>(K) "most essential" loop, fixing the important E and T
q w {F} r {K} y u i o p
a s d {T} g h j {E} l ; The "Tarmak1(Et)" transitional layout (T>F>E>K)
z x c v b n m
2) The G>(K)>N>(J) loop, bringing N into place
q w F r {G} y u i o p
a s d T {J} h {N} E l ; The "Tarmak2(EtN or EnT)" transitional layout (G>T>F>E>K>N>J)
z x c v b {K} m
Yes, it looks like a decent idea. The advantage, as you say, is getting T in place in step one which is a big boon for the stats in the first step; you also score the quite common F in the same fell swoop. This time you've avoided multiple misplacements, using K in step 1 and then J as before as the "odd man out". The number of keys moved is the same as the current Tarmak in all steps. This gets the benefits of your previous suggestions without the major disadvantages from the looks of it!
It'd work well, I'm sure. But I don't honestly think that many users will be stuck on step 1 for a long time, and then the consistency of only moving the J out of place seems like a decent compensation for the disadvantage of not getting T in place one step earlier. But different people will have different feelings about that, I'm sure.
I haven't compared bigram analyses, so I've no idea what the alternatives would look like in that department. But your step 1 would thrash mine in most analysis such as patorjk, I guess. I do believe the first step would test well against Minimak4, which is a lot of "test bang for the buck" but in my opinion has some flaws (changing the fingering of T, and moving the D for no great reason I can think of except not wanting to change its fingering). Could someone run a comparison of the current Tarmak1(En), the proposed Tarmak1(Et) and Minimak4 (E>K>T>D)? That'd be interesting.
Thinking about the Curl mod (which I do these days!), the displaced K makes it unattractive to implement a HK switch in step 1 as that'd lead to the common H being stuck in a bad temp position. So the most likely course of action would be to leave the T>F>E>K step as it is and do the whole Curl adaptation in step 2! But that makes this step rather largish so it might be split. At the least I'd recommend learning the Angle mod before the actual step then. The corresponding Tarmak2-Curl(Dvbg) step is simpler, obviously.
Tarmak(EtNROI)-Curl(DbgHk) said:2) The B>(K)>H>N>(J) loop, bringing N into place – with the Angle (V>C>X>Z>_) and Curl(DbgHk) mods
q w F r {B} y u i o p
a s d T g {K}{N} E l ; The "Tarmak2(EtN)-Curl(DbgHk)" transitional layout (V-_>B>T>F>E>K>H>N>J)
{z}{x}{c}{v}{J}{_}{H} m
I'm a little tempted to include your first step as an alternative. Then, people could choose which first step to use whether progressing to Curl or non-Curl. It'd have to be kind of snuck in as it were, as I don't want to change the main Tarmak now that it's well established – unless there are compelling arguments for it... :-)
[edit:]
So, I just had to run that test on Patorjk at least. It's an oversimplified analyzer, but I thought it'd be interesting nevertheless. Seems that engine really likes Minimak4 though, giving it nearly as much credit as Tarmak2! The test still shows that Tarmak1(Et) performs well.
Test: Patorjk – Most common English words
#1 Colemak (std.) 70.44
#2 Tarmak 2 (ET) 63.08
#3 Minimak 4-key 62.41
#4 Tarmak 1 (Et) 61.29
#5 Tarmak 1 (En) 60.65
Test: Patorjk – Alice In Wonderland, Chapter 1
#1 Colemak (std.) 65.83
#2 Tarmak 2 (ET) 61.92
#3 Minimak 4-key 61.87
#4 Tarmak 1 (Et) 60.73
#5 Tarmak 1 (En) 59.19
Last edited by DreymaR (05-Jun-2015 06:01:32)