Jon's motor is single phase, I assume Ben's must be a sort of 3 phase with that magnet to coil ratio.
I would have expected the single phase one with no skew to cog worse, but it does not have discrete magnets and the poles on that magnetised ceramic ring will not be well defined. The lower flux of the ceramic will cause far less cog.
All you can hope to do is make the attraction between the discrete magnets and those laminated poles average out over one rotation.
Changing the skew as Glenn described or making the edges of the polar projections less defined as I suggested should both work, but it will be experiment, no theory is going to get you to the exact point, I expect the makers attempt has removed 90% of the cog.
Ghurd has a point about trying to increase the rotor starting torque with lower speed, the starting losses do become a big issue with tiny machines if you need to start in low winds.
Flux
Flux