Ian Coley's approach is to prove $(I)$ and $(I) \implies (II) \implies (III)$
In proving $(I)$, how do we prove $$(-s)^2=s^2$$ without commutativity (but with $s=s^2$, if need be, and of course without $s+s=0$)?
Attempt 1:
$$-s=(-s)^2=(-s)(-s)\stackrel{(*)}{=}(-1)(s)(-1)(s)$$
$$s=(s)^2=(s)(s)$$
I'm stuck. Perhaps $-1$ commutes with every element of a ring assuming that the ring has a multiplicative identity $1$, but if the ring doesn't have a $1$, then I guess the ring has $-1$ only vacuously and therefore $(*)$ is meaningless.
Attempt 2:
$$-s=(-s)^2=(-s)(-s)$$
$$s=(s)^2=(s)(s)$$
$$\implies 0=s-s=s^2 + (-s)^2 \implies s^2 = -(-s)^2$$
I'm stuck.
Attempt 3:
$$-s=(-s)^2=(-s)(-s)$$
$$s=(s)^2=(s)(s)$$
$$\implies -s = -(s)^2=-(s)(s)$$
$$\implies (-s)(-s)=-(s)(s)$$
I'm stuck.