Ask HN: Why not compose repos together instead of using a Monorepo?

I have seen a lot of articles recently about Monorepos, attacking or defending them. From what I've read, the main benefits of monorepo are mainly:

- To share code between services

- To enforce code ownership and visibility

- To enforce standardization within the codebase

While the cons are that:

- It does not encourage isolation

- It requires a lot of tooling adaptation

- It's very hard to scale (VCS becomes very slow, you need a VFS, etc.)

My question for HN, is: If we used polyrepo, but with one main repository in which we would compose all the other repositories (as submodules for instance), wouldn't we get all the benefits of both worlds? Isolation AND homogenization? And all of that, just at the price of dealing with submodules?