I just remembered I have done the experiment Jack is talking about-there is one problem with repulsion and shielding you can't get around.
- when the shield is leaving from between two magnets in repulsion-the attractive force is double when the shield is about half way out-hence you gain nothing.
- if the magnets are not perfectly centered with the shield one magnet will repel the other magnet through the shield.
I have done this many times and cannot come out ahead, but I think this "C" shaped shield and the steorn configuration might work. It's always about millimeters, everything has to balance perfectly.It's a bitch in every sense of the word.

I have heard someone say that opposing magnets will keep their magnetisim without significantly weaking for thousands of years (at least about neodymium magnets). Is this true or not.

I'm not sure which one weakens the magnetic field, that should be determined IF you're looking for an attraction method, this may work.

There maybe some issues with my drawing regarding how harsh the cam should be.. but i'm bad with paint
its just a basic idea...
***EDIT***
The side view has an error, the LEMA w/ shield shown needs another magnet inside, (this is inline with the design on the patent, I just forgot to draw the other magnet)
TOP VIEW





SIDE VIEW

Honestly I do not know, It's a 50/50 split from what I have read, some say yes ,some no. On the quantum mechanics end they say yes, the magnets will weaken, the domains get disoriented over time, I tend to believe this theory more than the others,my neo's are getting weaker the more I experiment. They are 1" round x 1/2" thick N40 and I can tell they are getting weaker in just 6 months of playing around.

unshield when the rotating magnet is slightly passed the fixed magnet.

Pull shield perpendicular to rotation (not tangential) and the motor will run either direction

Helps to have fixed magnet larger than rotating magnets.

You must use opposing fields.

Demagnetization is a funstion of the distance between the magnets and the amount of demagnetizing force - all magnet mfg's (the large ones) give demag curves/data.  The shunt helps correct demagnetization - just like on the old horseshoe magnets.

Long magnets are more resistant to demag as are the rare-earth magnets.