Can’t believe this article doesn’t mention shaders. Node based programs to generate textures or the final shaders are everywhere, on all engines or render software. Unless you are writing very, very, very specific shaders, you can almost always solve it with a node flow and it pretty much is the norm.

Things that are highly parallelized and computation focused are the best for flow based programming, and shaders are probably the best example of that.

I occasionally use this style to work out problems with an element of distributed computation, and when I went to review the literature to do that again, this article came up. Often I end up with the final code being implemented conventionally, but it’s a helpful middle step to apply FBP concepts to work through “where” things are.

This seems ideal for something like ActivityPub where data is flowing between different people and places all the time. Looks like the tooling isn’t there yet, though.