For a while I had it in my mind that maybe you could save some memory while having safety just by using os.fork() after pre-importing a bunch of stuff; but then reference counting was pointed out to me and I realized it would be pointless -- the act of merely looking at an object changes the object, so copy-on-write means copy-everything.

Reference counting seems like it could be removed... except for all the C extensions that are specifically handling reference counting internally, removal of which I'm guessing would cause a lot of breakage. But then... it seems like those C extensions are exactly the problem, so if PyPy tries to keep them is it messing itself up? (Though I'm guessing it would somehow use proxy objects or some other trick to keep its internal object representation flexible.)