Hacker News Reader

Comment by jcoder | Hacker News Reader

If anyone is interested in comparing Colemak, Dvorak, and Qwerty for themselves, here's a small executable I wrote: https://github.com/bak/keyboard_battle. `gem install keyboard_battle`, then `keyboard_battle anytext.txt` or `keyboard_battle --bundled`. I believe it will work with Ruby 1.8, definitely on 1.9.

It compares both alternation (more is better) and reach (less is better). The main flaw is that it only counts ASCII characters, so may not help much for non-English texts.

View on HN

Do you have any summarized results to share from using this yourself?

Also, I might be interested in helping you get this to analyze non-ASCII texts.

Sure, here's an example of the output from `keyboard_battle --bundled`:

    texts/alice_underground.txt:
      qwerty:
        alternation_effort: 32118
        reach_effort: 51842
        raw_score: 83960
      dvorak:
        alternation_effort: 24476
        reach_effort: 30312
        raw_score: 54788
      colemak:
        alternation_effort: 29962
        reach_effort: 29403
        raw_score: 59365
    
    texts/declaration_of_independence.txt:
      qwerty:
        alternation_effort: 3049
        reach_effort: 5129
        raw_score: 8178
      dvorak:
        alternation_effort: 2237
        reach_effort: 2693
        raw_score: 4930
      colemak:
        alternation_effort: 2725
        reach_effort: 2469
        raw_score: 5194
    
    texts/gullivers.txt:
      qwerty:
        alternation_effort: 110812
        reach_effort: 181778
        raw_score: 292590
      dvorak:
        alternation_effort: 81836
        reach_effort: 103371
        raw_score: 185207
      colemak:
        alternation_effort: 100074
        reach_effort: 97327
        raw_score: 197401

The raw score is simply the sum of alternation and reach. Without research into whether alternation or reach matters more in the long run, I don't feel like I can provide an aggregate ranking.

Regardless, we can draw some rough conclusions like Dvorak is best on alternation, Colemak is best on reach, and they both blow Qwerty out of the water.

EDIT: Even though more alternation is better, the numbers above represent effort, meaning subsequent keypresses with the same hand, so lower numbers are always better.

As far as contributing goes, if it's possible to reliably support non-ASCII characters without special knowledge of the user's OS config, please take a crack at it (or at least open an issue). Thanks!

I wouldn't say alternation is always better -- in many cases, having successive keys close together is just as good if not better - "There" the first three letters "The" are typed with alternate hands, then the next three - "ere" are also easy to type fast because they're to each other.

I haven't used a dvorak keyboard, but the next to each other keys are really a part of my flow. I really don't know how "<esc>:wq", "<ctrl>a" and so on which are part of programming work would translate to dvorak.

I think you have a good point in that working in "command mode", or "chording", or whatever you want to call it, is different than writing prose. FWIW, I am a Vim user and I think it works great in Dvorak (https://news.ycombinator.com/item?id=4323873). `<ctrl>a` is the same as qwerty, `:wq` is typed on both hands, but as I say in that link, I prefer split hands for `hjkl`, and `wq` is pretty easy too.

To your point I would accept that in command mode, key position, mnemonics, etc. may be more important. In insert mode however I think it is, at the very least, a sound hypothesis that more alternation creates more comfort.

Reply on news.ycombinator.com