Let $A,B$ be commutative rings.
Defining a product of $B\otimes_{A}B$ as $(b_1 \otimes b_2)\cdot (b_3 \otimes b_4)=(b_1b_3)\otimes(b_2b_4)$, this becomes a commutative ring.
Defining $b\cdot(b_1 \otimes b_2)=(bb_1)\otimes b_2$ for $b,b_1,b_2 \in B$, this becomes a $B$-module.
Then, my textbook says, if I define as $b\cdot(b_1 \otimes b_2)=b_1\otimes (bb_2)$ instead, it defines a different $B$-module structure. I don't understand this.
Isn't the homomorphism defined by $\phi(b_1\otimes b_2)=b_2\otimes b_1$ an isomorphism of them?
Let $B$ be an $A$-algebra and $M$, $N$ be $B$-modules. Here is an example in which $M \otimes_A N$ and $N \otimes_A M$ are isomorphic as $A$-modules, but not as $B$-modules. $\mathbb Z[x]$ is a $\mathbb Z$-module via the inclusion $\mathbb Z \hookrightarrow \mathbb Z[x]$. Let $M = \mathbb Z[x]$ be a $\mathbb Z[x]$-module via the identity. Let $N = \mathbb Z[x]$ be a $\mathbb Z[x]$-module via evaluation at $0$. Then $M \otimes_\mathbb Z N \cong N \otimes_\mathbb Z M$ as $\mathbb Z$-modules. However, they aren't isomorphic as $\mathbb Z[x]$-modules, since one is annihilated by $x$ while the other isn't.