Yes, there are photos taken of the real thing. I guess what I'm missing is a video; it seems these days if it's not on YouTube it doesn't exist. And then even if it does the concept of 'originality' seems to be who can copy someone else's stuff to their contacts first!
But it does work. The decoupling mechanism, to separate the two axes of movement, is what took so long to work out (without the use of ball and socket joints) and actually works best because of a little twisting in the mechanism. The trick is there's an extra flexing point in the linkage on the left, that exactly lines up with the pivot point on the right. So up/down movement doesn't affect the left-hand-side; only right and left. On a strictly rigid system, when the stick is pushed over to the side, the up/down pivots would no longer align. But there's enough flexibility in the structure that it doesn't matter; and it's only on/off, not analogue, positioning that's important. So we get away with it. That decoupling mechanism is what everyone's copying, however their end results looks.
You'll find on a lot of analogue electronic joysticks there is no such uncoupling; one axial sensor will actually be mounted on the rotating shaft of the second axis. People building NXT joysticks usually use the same thing; you'll see one of those big position-sensing motors mounted so the whole thing can rotate. Mechanical aircraft controls are different still; a glider is the simplest where the joystick is mounted through a rotating shaft that handles left/right movement to control the ailerons. Forward/back operates a lever along that shaft that isn't properly 'de-rotated'; it's simply very long and has flexible end couplings, so the left/right movement hardly affects it. Though you could use some sort of sliding slip-ring to separate out the two axes in a shorter space.
Thanks for the voices of support, and I'm glad people have found it useful, anyway.