My understanding of when you should use constant or variable ratio is that its a judgement call on your part.
A film that is two hours long and is compressed (movie only) at constant ratio may show virtually no sign of having been compressed -> thus, use constant and be as happy as you have always been with DVD2One output.
A three hour long movie, on the other hand, has always been cause for debate about the quality of DVD2One output. Films like Gladiator or The Lord of the Rings reportedly show signs of quality deterioration when run through at constant ratio (as the older verions of DVD2One all used constant only). If you are one of those people who believed that DVD2ONE's quality was unacceptable on these long films, you now have the choice to use the variable option and allow more data to be allocated to the complex parts of a long film that would otherwise show deterioration.
I haven't actaully used the variable ratio option yet, but my understanding is that the quick scan of the disc that v1.20 does is meant to give the program an idea of where the most complex parts of the movie are, and thus allow for less compression of those areas at the expense of lower action sequences (or the credits).
Choose your own cutoff point and use constant for any films under a certain length, and variable for those over that length. Or, like I said above, if you have always been happy with every movie that you have backed up, don't worry about it and keep using constant.
Returning to the original question of whether variable or constant compression ratios are ideal, I have made a few observations. The variable compression setting does not work well with movie scenes with very little action or dark picture. During these scenes, the bitrate jumps around too much making the picture seem to come into and out of focus every few seconds. From my experience, it is best to use constant compression whenever possible, namely on shorter movies, and variable only on long movies. Has anyone else experienced a similar problem?