Of course it is arbitrary, just like every other moral compass out there ultimately is (as far as has been demonstrated). Practically, we already have a decent

Of course it is arbitrary, just like every other moral compass out there ultimately is (as far as has been demonstrated). Practically, we already have a decent idea what "wellbeing" and "suffering" mean. That means we can use them as our axiom even if the understanding is not perfect. If you waited for perfection, you'd be stuck doing nothing forever for the fear that it's immoral. We can still use science to examine where our axioms lead, and see if we still agree with the projected outcomes. Then we may have a reason to modify the arbitrary starting point.