No, the controls
definitely rotated, and I will prove it to you, in the mathematical sense of the word "prove":
Here, I've taken the three consoles
Smash has appeared on to an official degree. I've labelled what the buttons do, as well as their relation to one another - red shows the relation between the attack and special buttons, green between the special and upper jump button, blue between the attack and lower jump button, orange between the two jump buttons, and gray line's basically just there to try and match what you're seeing with your ">" layout.
Now, when I'm talking about how the buttons rotated, I'm focusing mainly on the green and red lines, because those are the ones I visualize. Notice how the layout is consistent for red and green for the first two, but rotated 90° counter-clockwise when the Classic Controller Pro enters the picture.
That is what I'm complaining about when I say things have rotated.
Meanwhile, you're visualizing the red and gray, probably spurred on by how the A button on the Gamecube controller is more in-line with the leftmost and rightmost buttons (as evidenced by the resulting quadrilateral being flatter on the bottom for the Gamecube than it is for the N64 or CC Pro). The red-gray relation is deliberately set up as you stated for the Gamecube controller and CC Pro, although I had to guess what sort of relationship that'd take for the N64 controller. In either case, it has to produce an acute angle with the red line, that always makes a right-facing arrow shape (">").
Now, you're also asserting, in the quoted post, that the controls never rotated. Let's assume that
is the case. Now, when the control layout started, that gray line more-or-less bisected the angle between the red and blue lines. If the controls did not rotate, it follows that it would still be bisecting this angle. However, setting up the relation between the red and gray lines
precisely as you argued it should be set up to maintain continuity between the Gamecube and CC Pro control schemes, you will note that instead, it becomes congruent with the blue line (it's hard to see in the image, but I put it just above the blue line), and does not even remotely bisect the angle created by the red/blue lines. Therefore, by proof by contradiction, the controls have demonstrably rotated,
Q.E.D..
Does this help illustrate where I'm coming from any better? I'm approaching this from the assumption that most controllers since the SNES are laid out in a quadrilateral formation, most typically a diamond with one button on the top and bottom and then two buttons in a middle row in between. This approach of mine extends even to the Gamecube's layout, even though it's stretched somewhat by the A button being more in-line with that aforementioned middle row (really, that resulting quadrilateral should probably be concave; I had to stretch it a bit to keep it convex like the other two). And when you approach it from that quadrilateral assumption, the controls rotated, as demonstrated. I believe people will be more likely to approach this from my approach than from the "these three buttons make a ">" shape" approach, since more controllers have that layout than do not.
