Scratch is overrated. The color blocks are a fantastic draw for getting a nonprofit funded, but for the children actually using it, the they're just an obstacle. Even at the youngest school ages, children grasp the grammatical concepts of programming effortlessly. It's the math that's the kicker--explaining modulus when they haven't gotten to division yet, binary & hex before they've fully grasped decimal, coordinate systems before they've learned negative numbers, etc...
IMHO, best modern tools for lern2code would be a tie between Lua/Love2d & p5.js.
And for 3rd place... you're going to think this is WAY OUT THERE... C--spare them the gcc esoterica of course--with sample project they can use as a scratchpad. Having types, but without all of the meta complexity of anonymous funcs, OOP, closures, dynamic typing, etc, takes a lot of the complexity (and frustration!) out of learning to program.
At the older & more social ages, the old web APIs were a wonderful stomping-ground. That avenue has since been narrowed-down & greatly so, for obvious reasons.
The Papert/LOGO/PARC stuff should be thrown into the fire. A military-industrial experiment on children; how quickly can we transform the random prole into a technician? Not that any of the researchers were "bad people" or even aware of it. It was a simple necessity of the times & typical "forest for the trees" situation. I think the numerical/visual/spatial primacy of this branch--trying to give kids a visual "feel" for mathematical representations before their brains are ready for the "real deal"--diverts focus away from many of the more "human" uses of the technology.
IMHO, best modern tools for lern2code would be a tie between Lua/Love2d & p5.js.
And for 3rd place... you're going to think this is WAY OUT THERE... C--spare them the gcc esoterica of course--with sample project they can use as a scratchpad. Having types, but without all of the meta complexity of anonymous funcs, OOP, closures, dynamic typing, etc, takes a lot of the complexity (and frustration!) out of learning to program.
At the older & more social ages, the old web APIs were a wonderful stomping-ground. That avenue has since been narrowed-down & greatly so, for obvious reasons.
The Papert/LOGO/PARC stuff should be thrown into the fire. A military-industrial experiment on children; how quickly can we transform the random prole into a technician? Not that any of the researchers were "bad people" or even aware of it. It was a simple necessity of the times & typical "forest for the trees" situation. I think the numerical/visual/spatial primacy of this branch--trying to give kids a visual "feel" for mathematical representations before their brains are ready for the "real deal"--diverts focus away from many of the more "human" uses of the technology.