The Norman layout [1] beats Workman at it's own metrics (or so it says on the tin), and from brief experimentation makes more sense to me.
I've always wanted to try an alternate layout, and this post led me to find minimak [2]. I'm typing with it right now and really enjoying the similarity to qwerty (only 4 keys changed) with reduced movement, plus 99% of my shortcuts remain the same.
I'm tempted to try out minimak. My only concern is that... do the numbers that get thrown around about percentage improvement only apply to touch typists?
I've never been strict enough with myself to learn touch typing (I still hit 120WPM so speed isn't an issue) and worry that the benefit won't be as obvious.
The metric I lean on the most is the one published by Andong at http://www.andong.co.uk/dvorak/Default.aspx. I guess that's odd since I spent the least amount of time talking about it and put it at the bottom of the page. It is focused on metrics of efficiency and effort. If you wanted speed, you might want to look at things like finger distance and home row usage more closely, in which case Arensito or Colemak might have an edge.
I must confess, since I swapped R and H in revision 2, Workman usually gets the edge on Norman using Workman metrics, depending on the input text. I guess I'll hunt down whatever text revisions are appropriate on normanlayout.info.
I've always wanted to try an alternate layout, and this post led me to find minimak [2]. I'm typing with it right now and really enjoying the similarity to qwerty (only 4 keys changed) with reduced movement, plus 99% of my shortcuts remain the same.
[1] http://normanlayout.info
[2] http://minimak.org