> to create a powerful new tool in cryptography.

What is that new powerful tool in cryptography, then?

> He wanted to build zero-knowledge proofs that weren’t interactive. Thirty years earlier, Goldreich and Oren had established that such proofs are impossible.

I'm not sure what "interactive" means here, but I thought ZK-SNARKs were already non-interactive.

It seems the article has nothing to do with anything practical..

Is the approach analogous to one way hash? But with mathematical statements?

Given that they can’t be proven, so it’s effectively unpredictable and “un-generatable” ?

How is this not security through obscurity?