Sorry if this is going too far off topic, but just to illustrate the phenomenon I have made up a graphic:
Now, assuming the animation has loaded completely and runs properly, what you should see in the first panel is Superfrog gliding along smoothly at even intervals, two "pixels" or so at a time. This is what the original 50 Hz output should look like (obviously slowed down a lot here).
The second panel tries to illustrate what happens when the output is 60 Hz instead. As the frames have to be sent out quicker, every now and then there is nothing to send, as a new frame has not come in yet. Therefore the current frame has to be sent out again, and this leads to small stops or stutter in the movement. This is most likely what all of these scan converter boxes, Indivisions etc. do in the situation.
Another option sometimes used in PAL-NTSC video conversions is to blend the frames in the appropriate ratio. So for example if the output frame is best represented by 30% of the previous frame and 70% of the current frame, such a blend is created. This can look decent in the case of television, but the ghosting is pretty obvious if you look closer, particularly with sharp content such as video games.
The "holy grail" of standards conversion would be perfect motion interpolation of the content. 100/120 Hz TV's with "pure motion" etc. technology sort of do this, they estimate an in-between frame to simulate more fluid motion. While this can work reasonably well, it's much more difficult to make an arbitrary conversion. In a 50->60 conversion most output frames would have to be synthetic ones, rather than just every other one. Also even the best algorithms tend to fall apart when there is complex motion, such as things passing in front of each other and so on.
Now again, for anyone who is not bothered by such inaccuracies in motion, this does not matter. But I'm pretty sure the problem itself has not been overcome in any device yet.