I've recently been testing out the capabilities of the script node. I've known for some time it's been in the works. Now that it's finally released it has been well worth the wait. The key thing for me is the ability to condense logic into a few lines of code. Code is just way more efficient for more complex tasks than connecting pre-defined nodes. Plus it helps to try avoiding the Deutsch limit (https://en.wikipedia.org/wiki/Deutsch_limit)

Looping is key too. So much of visual programming is about data mapping. Having the ability to create arbitary loops is key. There are many cases where looping makes the task so much more managable. I took an example of iteratively dividing a polygon into smaller and smaller subsets. It was pretty easy to implement as a script node. I'm hoping, as I go further into it, I'll manage to find ways to condense the logic even more.

I'd love to hear comments from anyone else that's been testing this new capability.