This is the layout of Nanaju:
I saw several keyboard layouts developed with the "waterfall model" style. Each of those layouts was derived from a set of principles. For example, "Bigrams should not be typed with adjacent fingers" is one of such principles of Dvorak keyboard. Sadly this principle is wrong after electronic typewriters. The principle is correct if the typewriter inhibits key rollover mechanically. Now we cannot live without diodes for good key rollover.
Many of Dvorak keyboard principles are still relevant today. But the set of principles is not. The set was short-lived compared to humans. Moreover I have seen some absurd principles in other places. Dvorak keyboard doesn't have such absurd principles and this is why it has been successful comapred to its rivals. But how can we avoid contamination of absurd principles? Just by testing in the real world, no other way. I am very skeptical about the effort to find the "best" layout from a set of principles. I already tried it in Japanese kana layout. My thoughts: It is a good way to find flaws in the set of principles, not to find a fruitful result.
I did not try to find the "best" layout for Nanaju. Instead of the best derived from something untested, I pursued something tested by making/testing many, more than ten prototypes in the haphazard way on a patchwork basis. I pushed myself until I groaned "Enough, it's enough. I did my best. I don't believe anymore that the next prototype may be fruitful. I already have spent enough energy for making/testing prototypes". Of course, the next prototype can hit a jackpot. I cannot prove the inexistence or unreachability of a jackpot because I don't have any methodology. I don't care. I am not an academic nor a sales pitch writer.
So I can tell you why N-th prototype layout isn't as good as the final product Nanaju layout, but I cannot tell you why the arrow keys are there. If I had to say, I was burned out before finding better places for them.
Although I believe that I have found some rules.
Rule one: You should not place navigation keys (arrows, PageUp/Down, Home, and End) on left and right hand both. To select text, it’s common action to input Shift + End first and then Shift + arrow. It occurs while selecting from the cursor to the end of the line except the period, for example. If you place arrow keys on the right hand and Home / End keys on the left hand, in many cases, the layout will force you to release Shift key after Shift + End. Why? At the moment, you will be pressing Shift key for Shift + End by right hand and feel difficult to keep pressing it for Shift + arrow because the arrow key is on the same right hand. This is cumbersome and occurs quite often.
Rule two: The sharp corners (Esc, Del, Win, PageUp, and Backspace) are not good places for typing because it is distant. But they are good places for hitting because they have wider margin than usual places. In typing, you type a series of letter keystrokes without waiting for feedback. In hitting, you may hit the key repeatedly without waiting for feedback, but the action is separated from a series of letter keystrokes by waiting for feedback. The separation makes the distance unnoticeable.
About patchwork, look at the Del key. It came from the RUB OUT key of the Space-cadet keyboard. There are quite a lot of layouts which have centered Shift key. "英数" and "かな" keys are stolen from Microsoft Surface Japanese version (They are irrelevant keys for English. Please remap by Vial). In this area, the stories about abandoned ideas will be much more valuable than adopted ideas, but they are long long stories...
