While his article makes some interesting discussion, and some nice points, the most important piece of his 'persuasion' is in the experimental evidence he is seeking to provide. That evidence comes from a study conducted through the Navy and he describes how in their study Dvorak did not display the 'benefit' let alone a significant one.
However there is an enourmous flaw that I can't ignore, the complete lack of validity to this study. Having spent time studying it (see next paragraph) that 'study' is basically a text book example of how NOT to perform experimental research. It violated just about every single method of control that it encountered and was extremely biased. So much so that it was later used as a case study of how not to conduct research in one of my classes.
I say this all because I am familiar with the topic. When I was in college (graduated in '06) I took two courses of note: Human Factors in Design, and Research Methods and Statistics. The professor who teaches them both, had (and usually has) many students cross-enrolled and allows the semester project to be combined if appropriate subject matter is used. I combined my projects and the focus area was Dvorak and alternative layouts over QWERTY.
What I uncovered in background in a nutshell was as I said above. A seriously flawed study. Inadequate rest was given to the subjects, no control over ordering, while it can't be blind to subject the ordering the subjects were using should have been blind to the proctors. The assumed acceptable time for training on DVORAK was an extremely lowball guess. Furthermore there was a study the Navy 'funded' that showed the results the other way which they abandoned (as well as other studies from other branches). The entire study couldn't have been much more biased.
Furthermore if you are familiar with QWERTY, the very DESIGN of it was to slow you down. QWERTY was designed to prevent typewriters from jamming. Jams were caused by people typing too rapidly; QWERTY was enough of a hinderence to prevent it. By the time computers rolled around, there were so many typewriters and people trained and invested in the QWERTY layout that people simply didn't want to switch. Additionally companies likely didn't (and don't) want to replace their entire office at once; or to have to spend time and effort re-training.
We can see this kind of legacy hinderence throughout the market place; where despite a better product being around, the adoption of prior less ideal but acceptable product is so widespread that it can be disadvantageous to switch.
I'm not even going to waste my time talking about the details of the ergonomics. The only point I will make there is one that even DVORAK still had room for improvement on. That is balance of finger use. On a QWERTY keyboard there is an extreme overuse of the pinky and ring fingers compared to the middle, index, and sadly the most dexterous of all the thumb; (especially as compared to their relative strengths)
The problem is that while for a machine pushing each key is equally difficult; and thus jams are matter of angles; it is not so for a human. Pushing certain keys is more difficult than others.
I apologize that much of this has been said prolly, but I couldn't avoid commenting on a horrible post. Take a 13 year old article now, and make that article about a then 19 year old experiment. Clearly in the past 32 years or so we haven't had any advances in the standards for conducting valid research. Let alone advances in our very understanding of the underlying factors (many of which didn't have names when this study was conducted).
FAIL :-)