and by moving the common digraphs further apart it also slows down the typing - especially if both letters are on the same half of the board - and therefore most likely used by the same hand.
Slowing down the typing not only stops adjacent bars from binding, it gives the bars time to return to rest - and thus get out of the way of the next.
That is no longer necessary (in fact, it was no longer necessary within Sholes' lifetime as the quality of machines improved - but that did not prompt him to change the layout to one that made more sense) and QWERTY users are stuck with a large number of combinations of letters that require hurdling over the home keys or using the same finger to hit successive letters or both - for example "ce" requires the left middle finger to reach down to the bottom row then hurdle the home row to the top row.
On Dvorak, that same digraph is a single reach one row up with one finger and a home key with another finger (on the other hand, even) - equates to pressing "ID" on a QWERTY board.
Which can you touch type fastest on a QWERTY board: "CE" (left middle down then hurdle home key (D) to top row) or "ID" (Right middle up then press left middle)? Try typing both those as fast as possible and see which you can do faster.
The combination "kjd" is utter nonsense on the QWERTY board, but it is very easy and quick to type - all home letters, no fingers need to move anywhere but straight down on their respective caps. Right middle then index, rapidly followed by left middle.
On the Dvorak layout, those three keys correspond to the letters "t", "h" and "e" so that bit of easy-to-type "nonsense" on the QWERTY board is one of the most common words in the English language if the keys are arranged in the Dvorak layout.
Each of those three letters require a reach on the QWERTY layout.
"Minimum": none of the keys are on QWERTY's home row, index finger has to switch between m and n on the bottom row as well as hurdle the home row to reach u on the top.
On Dvorak, the only letter not on the home row is "M" - and the index finger can hover on it while typing that word - just as the left index can hover over the "I" in readiness for the second one before returning to its home over the "U".
Also, on the QWERTY board, "minimum" is typed solely with the right hand. On Dvorak it is spread between the hands.
Dvorak and others based their designs on watching touch typists at work then striving to minimise muscle movements for comfort and efficiency.
Dvorak and his brother in law studied slow motion footage of typists, the common letters of the English language and the physiology of the hand and based their design on fairly sound ideas such as it being faster and easier if you don't have to move your fingers from the home row.
Given that QWERTY only has one vowel on the home row, typing most words entails taking your fingers from the home row - automatically slowing you down and increasing fatigue.
Other factors come into it - alternating hands, progression from outer to inner fingers, reaching up a row easier and faster than curling down a row and so on.
It's been calculated that a QWERTY typist's fingers move around twenty times further than that of a Dvorak typist's to type the same text - I'm sure I can walk a mile in less time than I can walk 20.
Without taking the time or effort to reach away from my 8 home keys, I can type nearly 1500 words. On the QWERTY board the same keys give less than 160 words. That's around 1340 words where a QWERTY typist has to reach for keys while I do not.
If I am typing those words at 35wpm and not moving my fingers from the home key and a QWERTY typist is typing the same words, moving all about over the board at the same wpm, then obviously their fingers are moving physically faster than mine. If they are making 70 to 100 wpm, then they are moving their fingers a lot faster than I am.
Now, how fast would those typists be if they didn't move their fingers from the home keys at all? How fast would they be if instead of reaching, they just pressed as fast as they could - what would their characters-per-minute rate get up to?
The problem with saying "people who have retrained in Dvorak aren't much faster than they were with QWERTY" is that it's just based on characters per minute, they are not showing any metric on how fast they are actually moving their fingers whilst doing it.
While their typing speed may be the same or a trifle faster, they will be actually moving their fingers less distance and, therefore, not as fast as they had to do so before to type the same common words in the same amount of time.
Now, there are a variety of factors in that:
One, how fast can the person process the stuff they want/need to type? If they lose the plot trying to formulate something faster than 100 words a minute, that is going to limit their typing speed no matter what layout they are using.
Also, if you type 120 per minute on QWERTY and move to Dvorak, are you really going to go all out to increase your speed to 200wpm or even 150wpm if you can type 125-130wpm with less effort than it previously took you to type 120?
You might have the potential to type like a demon on Dvorak but in all honesty, if you can match or slightly better your speed and decrease your effort by a significant amount in the process, you're not going to bother trying to exceed your typing speed by fantastic amounts.
You might be able to run the mile in under four minutes but if your job only requires you to do it in half an hour, you'll do it in 25 minutes just to look good without overdoing it.
Unless there's some incentive to crank up the fingers to the same blinding speed as before, most people are not going to bother, they will do just enough to meet the requirements of their task.