From the technical POV a scandoubler stores a single scan line and doubles the horizontal scan rate by outputing the line twice - this doesn't necessarily work with interlaced modes.
In contrast, a flicker fixer stores an entire video field (half of an interlaced frame) and alternately outputs a line from the previous field and one of the current. This nicely converts interlaced screens to progressive but produces ghost artifacts for moving objects (since half of the frame is 20 ms older than the rest).
