You're absolutely correct. Allow me to replace it by the following, clocking in at 87.

perl -e'sub h{pop=~/(...)(.)/&&{M,1e3,G,1e6}->{$2}+$1}print sort{h($b)
+<=>h$a}`du -h`'
[download]

update: No wait! There's something in your idea of bitanding... Just have to get rid of the second spaceship comparator...

perl -e'sub c{pop=~/[GMK]/;(ord$&&7).1e3+$`}print sort{c($b)<=>c$a}`du
+ -h`'
[download]

76 strokes.

Alas, your last two methods put a 400K file before a 399M file... For the first one, I added a "K" key to your anonymous hash:

perl -e 'sub h{pop=~/^([\d.]+)(.)/&&{K,1e1,M,1e3,G,1e6}->{$2}+$1}print
+ sort{h($b)<=>h$a}`du -h`'
[download]

For the second one, I think you're in a bind because ord(k)&7 and ord(K)&7 are equal...

And I offer another method, bumming heavily from yours, longer but possibly faster on large filesystems:

perl -e 'print map substr($_,8),reverse sort map sprintf("%8d",(/^([\d
+.]+)([kKMG])/)?{K,1e1,M,1e3,G,1e6}->{$2}+$1:$1).$_,`du -h`'
[download]

Oh, I see. Could I then also change the regex to save two more characters?

perl -e'sub c{pop=~/.\s/;(ord$&&7).1e3+$`}print sort{c($b)<=>c$a}`du -
+h`'
[download]

74 characters.

And, in fact, since we are only using .1e3 to make sure the sizes sort properly, why not save one more:

perl -e'sub c{pop=~/.\s/;(ord$&&7)**7+$`}print sort{c($b)<=>c$a}`du -h
+`'
[download]

73 characters.

Slick! But how about 70, using 'x' instead of '**', and 'die' instead of 'print':

perl -e'sub c{pop=~/.\s/;(ord$&&7)x7+$`}die sort{c($b)<=>c$a}`du -h`'
[download]