I guess the idea is to mimic the increased cognitive load of dyslexia to give non-dyslexic people a feel for it, without making each word as difficult as deciphering full anagrams.
Some of the words pop out to me immediately, but others I have to stare at for a while before I can figure out what the word is. The way the letters jump around puts an immediate stress on my brain that is really unpleasant.
From what I know of dyslexia, wouldn't it be more appropriate to replace each letter with a weighted random selection from groupings of similar glyphs?
a e g 6 9 @
b d p q P
c n u C G U ( )
f t k x H K X +
h y F T Y
i ! : ;
j r J L 7 [ ]
l I 1 / | \
m w E 3 { }
o D O Q 0
s z S Z 2 5 ? $
v A V ^ 4 < >
B R 8 &
M N W
. , ' `
* " ~
# =
%
_ -
There are also Unicode glyphs that resemble English letters turned upside down or sideways.Some psychology researchers can really mess with your brain by aiming a camera at your eye and changing the letters of a text during a saccade. When your eyes are moving to a new fixation point, you become temporarily blind to changes in what you are looking at for a fraction of a second.
Obviously, that can't be done in a webpage demonstration without some fool giving control of their camera to your javascript. Then you would have to calibrate the gaze detection. And your program would have to run fast enough to update during a saccade.