*already* computers generate bytecode that's
   improbable/impossible for humans to decode. Computers have been
   that way from the day of the very first compiler in the early
   1950s. [thanks Grace Hopper! awesome lady she was]

   So for a computer to be incomprehensible to a human is already
   here.

   Now, can we design a computer more intelligent than us?

   That's where it's tricky.

   _what's the measuring stick_?

   _How_ can we measure the difference between "incomprehensible"
   and "super-intelligent" - if we're incapable of being _more
   intelligent_?

   We have to be at a higher level to measure at a lower level.
   Even when we measure bigger things, it's thanks to the power of
   Optiks that things are measurable:

   How else does the Universe fit in a 2mm circle in front of our
   heads?

   Even if we designed a chain of less and less comprehendable
   computers going *upwards* to reach the "super-intelligent" and
   have them back-translate DOWN to our level... we have no way of
   verifying their correctness.

   No way to measure. No way to know.