Mathematically speaking, the blockchain is a DAG - just a very simple one

(blockchain depicted as a DAG)

A more generic DAG looks like this:

So yes, of course it is worth exploring and it's being explored by many projects.

But the "difference" you spot is not one. Depending on what you want to do or search for, it is actually much more complex to search a generic, unconstrained DAG than a much simplified and streamlined one that is better known as a "blockchain"

You can do a lot of things with a DAG. Back in 2002, I implemented a kNN search algorithm on a DAG that was an ontology of the medical domain: each "node" of the graph was not containing a "state" of the state machine like in cryptocurrency projects but instead was a concept from the medical domain such as for instance "illness" or "leg". The "child-parent" relationships that were linking the nodes of the graph were not transactions nor hashes of the previous "state snapshot" but ontological relationships such as "isPartOf" or "isSubclassOf"

That shows the versatility of DAGs

However one thing I'm skeptical about is the ability of "multidimensional DAGs" (non-blockchain DAGs) to implement a ledger. Maybe there are solutions, but they are not readily apparent