Once, a student wondered about many things, and questioned constantly in his quest to attain enlightenment. One day, he asked the question "who and what am I, so that I can seek the path to enlightenment?"
No one answered, and in that moment the student was enlightened.
I often wonder who and what I am, and in the wondering hope to move my experiential state vector towards a more optimal phase where information processed gives the greatest possible return on processing energy, thus increasing total wisdom. On occasion, I wonder if this wondering is not itself enlightenment of a kind. Consider this koan:
"A Zen master named Gisan asked a young student to bring him a pail of water to cool his bath.
The student brought the water and, after cooling the bath, threw on to the ground the little that was left over.
"You dunce!" the master scolded him. "Why didn't you give the rest of the water to the plants? What right have you to waste even a drop of water in this temple?"
Clearly, the student already knew all that he needed to become enlightened, since his master did not provide him any new learning, nor pose a question the answering of which would lead to new learning, of anything outside himself. Merely, he gave the student a cause to reflect upon himself ("What right?") and his actions ("Why?"), and in so doing learn what it was that had previously seperated him from his Zen state.
Compassion draws the enlightened one back, to continue to exist in disharmony with everyone unenlightened. Compassion is empathy. Empathy is fellow feeling. It is all predicated on knowledge of one's fellow, which is the encapsulation of the existence of other minds within a theory of mind.
In Accelerando, Charles Stross talks about the predictability of a normal human to one whose cognitive functions have accelerated, i.e. been boosted by direct interface with technology (something another TED'er - the insufferable but farsighted Ray Kurzweil - talks about). The theory of mind becomes so powerful that the normal human is essentially encapsulable as a predictive model (theories are postulated to predict real phenomena).
Can one feel compassion for an automata? A completely predictable person might look as such to the modeller. But the knowledge of the person required to build the model is itself an emotive context, and processing affective information causes affect. So compassion arises from knowledge, even when that knowledge is great as to be 'weakly godlike'.