you see you fight very soon against the rest of the world.
This should not be your aim, they are wrong.
Concentrate on that what you already have found.
I am more than happy with this, fighting against the rest of the world can be very constructive, when it is done politely and respectfully.
I agree that they a wrong, i just don't know what it is exactly they are wrong about. It is very clear that they are not wrong about everything.
I am concentrating on what i have aready found. The problem is it is very slow, i do not have access to machinery and reproducing many parts with hand tools is very time consuming and the lack of precision only increases the complexity. When everything falls precisely where it should there is no need for additional guides and deflectors to aline the different mechanismes.
We are both trying to do the same thing, i agree. There is however, in my opinion one big difference that i think very important.
Your octagon exists. And it is it's existance that is the problem. It is always having an effect on the wheel and the effect it has on the wheel is balancing out. If you could create an imaginary octagon that does exactly what your real octagon does you would not have this problem.
Imagine a rope that is fixed to the rim of your wheel that does one loop around the octagon. You have two circles one inside the other sitting at the bottom. The rope crosses itself where they touch. When you rotate the wheel with the octagon doing exactly what you want it to do, observe the rope (fix an imaginary weight to it). Observe the path of the weight for a few turns. If you can gat the weights to follow that path you do not need the octagon because the weights will be creating the same COM.
You will notice that the length of the rope is smaller going around the octagon than going around the rim. Therefore when the weight has gone around the octagon it cannot go around the wheel at the same place as where it left the wheel.
The octagon rotates more than 360° with every full turn of the wheel. This is exactly what you are obseving with your rocking/tilting octagon.
You have already observed that the octagon wants to tilt/rock all of it's own. This happens too late, this is the problem. You need to either accelerate it vertically so that it falls to the side or accelerate it horizantally so that it is always off center.
The swinging weights that follow the path of the rope (when going around the octagon) also want to swing naturally. The swinging of each weight needs to be accelerated horizantally a fraction to get the continual off center you are trying to get by moving a real octagon.
I think the horse of the cart is the weight that takes the rim of the wheel over the 12 oclock while pulling the weight that crosses the wheel at the center. Or in your case, pulling the weight at the top of the octagon.
When the cart meets the rim it then swaps to become a horse and pulls the horse (that has swapped to become a cart). This will give the exact same path for each and every weight as you have observed with the imaginary weight on the rope looped around the octagon.