I've noticed that most visual programming projects seem to be focused on new (and often unintentionally more difficult) ways to build traditional difficult to use end user apps, instead of making frameworks that help make end user apps themselves easier to use. Might it be more productive to work backwards from making life better for the end user as the path towards better developer tools? e.g. "We want end users to be able to do X, Y, Z in all situations, so what tools do we make to get there?"

"in all situations" = interop with every program the end user could possibly run? Or every possible task someone might want to perform?

@Kartik Agaram kinda yeah :). Reading "The Humane Interface" or whatever by Jef Raskin was influential to me in that regard, as a design at roughly that scope. Another angle: it's beneficial to users if the "programming" interface of their system closely resembles how you do things "by hand". You can see this in effect today with some power-user oriented programs where there's a blurry boundary between programming and high-leverage manual interactions (I'm thinking kakoune, but Vim will do, I suspect Photoshop too). Leaning into that idea tightly ties your programming and UI models.

didn't frontpage and dreamweaver take this approach?

I'm not familiar with Frontpage or Dreamweaver, but I see HyperCard as taking this approach. e.g. "Every user should be able to extend this in certain ways, and we're going to accomplish that though some abstraction and uniformity."

That helps a lot. I wonder if you'll find my new project to fit the bill: https://github.com/akkartik/teliva

Interesting. Automator, which adds simple scripting to Figma, would perhaps be something that answers this in the domain of a design tool. Of course, similar functionality exists also in Photoshop, Illustrator, etc to some extent. https://automator.design