To be more specific, if I imagine to work with isolated transactions chained to each other according to money flow, then I understand that we wish to have a single chain because forks would mean the presence of double-spend attempts.
But blocks actually are an aggregate of transactions happened in a certain window of time, so I find hard to understand the necessity of not having forks in a chain of blocks. Are transaction chains and block chains two very different things with different purposes? What I understood is that the first has the goal to order temporally the transactions while the second serves to make tamper proof the ledger by incorporating hashes of a block in another one.
In my mental model the time ordering goal is already reached by the transactions data structure itself while the chain of blocks has the (only) purpose of making tamper proof the structure. What am I not understanding?