The quote goes something like this:
There's no problem in Computer Science that can't be solved by adding another layer of abstraction to it
(Copied in this wording from http://blogs.oracle.com/fcmartin/2009/01/pardon_my_dust.html)
There are a number of variations but I have been unable to find an originator. Since I quite like the quote with its tongue in cheek and quite a bit of truth, I would be interested to hear if anybody knows where this may come from.
This website attributes it to David Wheeler, who worked on EDSAC among his other early accomplishments. His Wikipedia bio also suggests he might have originated it. This is probably your best bet.
The form they give is:
Any problem in computer science can be solved with another layer of indirection. But that usually will create another problem
I first heard it when I was a CS freshman in '85 while our instructor was introducing pointers in our CS101 course. I got the impression it was a canard even then.
If anyone is looking to attribute it to someone, it had better be someone at least moderately famous back in '85. That probably rules out Koenig.