float ballX;
float ballY;
float ballR;
float dX;
float dY;
float paddleX;
float paddleY;
float paddleW;
float paddleH;

void reset() {
  ballX = 20;
  ballY = 20;
  ballR = 10;
  dX = random(1, 2);
  dY = random(1, 2);
  paddleX = width - 15;
  paddleY = 10;
  paddleW = 10;
  paddleH = 30;
}

void setup() {
  size(400, 400);
  reset();
}

and add a new conditional in the keyPressed method to check whether the reset key (I’ve chosen the R key here) has been pressed:

// based on code from http://processing.org/reference/keyCode.html
void keyPressed() {
  if (key == 'r' || key == 'R') {
    reset();
  }
  if (key == CODED) {
    if (keyCode == UP) {
      paddleY = paddleY - paddleH;
    }
    else if (keyCode == DOWN) {
      paddleY = paddleY + paddleH;
    }
  }
}

void mouseMoved() {
  paddleY = mouseY;
}

1. Begin by learning to read and explain code. Children don’t learn their native language by beginning with writing short stories; they learn to hear it, speak it, read it, and only then to write it themselves. Programmers, on the other hand, are handed syntactic formalisms of the language of interest and expected to work from that point.

2. Focus on the meaning of code, rather than the syntax, initially. Too often, when novices are asked to explain what code does, they respond in a purely functional way (“well, we assign 2 to x, then…”), rather than looking at the big picture, i.e., what is the intention that the code represents?

3. Work with code that can be refactored transparently. (See my post Goodbye, Hello World? as an example.) One needs to see the same problem context repeatedly, improved and expanded at each step, so that one can appreciate how the changes being made reflect improvements in this particular solution to the problem. Students need to see more than just single snapshots of code; they need to see code evolve as ones understanding of the problem changes.

4. Work on problems that Matter (in the societal sense), or at least problems that matter (to the individual). The popular media computation curriculum from Mark Guzdial and Barbara Ericson attempts to address this by working with digital media files (images, sounds, video). Mike Buckley at University Buffalo engages his students with the needs of real people via a “Socially Relevant Computing” curriculum. I think there’s room to blend the two.

5. Remember that we can’t solve problems that students haven’t yet experienced and learned to value. I can teach the mechanics of matrix algebra to a 4th grade student, but it’s a solution to problems that the 4th grader doesn’t yet have. Because of this (and possibly other issues regarding the developmental appropriateness of the topic), the learning won’t be “sticky”. In other words, we need to begin with questions, so that the answers are meaningful.

The best we can do, as you say, is to make our values explicit. This would foster both fairness and efficiency, cutting out unnecessary mysteries about what one is in for in an educational program.

The relationship of explicitness to forewarning about cognitive overload is something that I had never thought of. Excellent idea!