Fark linked up this interesting article describing how fish select their cleaners:

The researchers set up an experiment where client fish could observe two cleaners, one dutifully munching away at another client's parasites and another swimming freely near another client.

Taking the cleaner fishes' past performances into account, the clients chose the cooperative, hard-working fish significantly more often than the loafer.

So maybe our goldfish really aren't (yet more) self-propelled eating machines coming up with clever ways to widdle in our house. They may be smart self-propelled eating machines coming up with clever ways to widdle in our house.