I got interested in steno once, and even designed and started making my own custom open source steno keyboard before Open Steno existed (though that's a story for another day). This led me to look in to learning steno, and I discovered that the main reason that stenographers type as fast as they do is that they memorize tons of macros (in vim and emacs terms). The triggering of a macro will result in a whole phrase, or even sentence being written out (or just a shorthand for that word/phrase is written out, and that can later be converted by hand in to that word/phrase).

On an ascii keyboard, that would be like typing something like \d and having the phrase "The defendant states that" appear.

It's obvious that simply by using that technique alone (which can be used on a regular keyboard too), one can type many orders of magnitude faster than someone who's typing out each individual word letter by letter (or even phonetically). Some steno systems will give the stenographer hundreds if not thousands of such macros to memorize.

Of course, there's also the phonetic spelling stenographers use if they don't already know a macro for a given word (or for a proper noun), and there's the funky chorded keyboard that makes it efficient to type thousands of different macros with a relatively few keys and without moving the fingers much. But that's secondary to the heavy reliance on macros.

I hate memorization, and didn't want to spend the years that it would take to master steno, so never really wound up pursuing this myself. So I don't actually know steno myself, so please take the above explanation with a grain or two of salt. But that's my understanding of how the greatest speed advantages are achieved with steno.

I'd love to hear from some professional stenographers as to whether my impression was correct.