I'm not sure what you mean when you say "
this", but here's my interpretation of what you mean. Feel free to clarify/reply; I check thread replies on HN every couple days, hopefully others chime in too.
The most popular methods to teach language put an incredibly strong emphasis on the link between visual glyphs and the sounds they're associated with. This technique seems highly practical on the surface because it aligns reading and talking, so effort in one area leads to reward in the other.
This system isn't perfect (or IMO acceptable/sane), though, because by design, it doesn't teach/provide an "off-switch" for when you want to do one without the other; rather, it conveys the opposite idea that one always implies the other. So either you intuitively figure out the nuances of where and how to differentiate visual and audial yourself, or you end up learning there are kids in the classroom who who can't read silently because they "don't get" (read: weren't taught) how to do that.
And then we crash into the whole maximum reading thing speed a few years later, which is so subtle and innate that most people don't even realize they have the issue at all. Just like with silently reading, a lucky few retain enough intuition to keep the visual and audial side of things completely separate in spite of how they were (mis)taught: one of my friends has the ability to look at (ie, read) his entire laptop screen as a single whole. (Of course I've signed up for lessons :P but it remains to be seen whether I can learn... that.)
But that's what speedreading is: reading purely visually, not with the clunky, time-based words->sounds->comprehension thing.
Quite possibly, as you read this, you're hearing it in your speaking voice or a mental voice. Not only is that capability innately time-based, rendering accent, tone and pause use mental resources.
With effort, you can consistently turn this off. It's just really really hard to learn to, 1) because we've been doing it all our lives and 2) because it was ingrained so early.
I understand that comprehension speed can go through the roof once this is done, because using the visual parts of the brain to process reading directly - ie, skipping glyphs->sounds->comprehension and going straight to glyphs->comprehension - is basically the way to do it. Sadly, only very few people to this.
The read-slow technique is a new one to me, but it seems to work, sort of. A better method I've also read about is to bog-down the sound processing parts of the brain with an activity that completely occupies all sound processing: repeating a sequence of letters over and over, like "A B C A B C A B C A B C ..." while you read. The visual and audial parts of the brain both ultimately do reach the comprehension parts, but the audial parts have the most connections. Bogging down audial processing with a task like the one above means you fall back on the visual->comprehension connections... which is what you want.
Rebalancing everything so comprehension is predominantly visual... yeah, that would definitely the subject of some efficiency research, because without some kind of secret-sauce learning method I'd imagine it would take years.