It’s rather problematic to go back only so far as 1972, I feel. So far as I know, the most recent election where no county voted for the opposite party for the last time as of 2017 was 1920 – and I have not checked that one perfectly. There are, too, several in 1916 (e.g. Lawrence County, South Dakota) in Republican strongholds of the Great Basin and Great Plains and in 1912 (e.g. Pulaski County, Kentucky and Hamilton County, Indiana) over parts of Appalachia and areas to the west. 1968 and 1960 have only a few – but with 1960 there are some of special note like Multnomah County, Oregon (anti-Catholicism perhaps), Johnson County, Iowa and on the opposite side several contiguous Gulf Coast counties in Alabama and North Florida.
I wonder if the past few elections (say, since 2000 or 2004) could be grouped together to separate those counties going back before 1972? It’s of more interest to me, and perhaps to many psephologists, to look at the deep roots of modern voting patterns than at relatively recent changes.
Another alternative would be to use different colours, as opposed to merely different shades of purple, to solve both problems of not going back as far as one would desire and of poor colour resolution.
I did an
alternate one that goes back to 1960, but the primary problem one runs into when building such a map is the number of elections/color gradient. As you can see with the 1960 map, even counties that haven't flipped in 30 years don't look as if it's been that long ago (relative in shading to recent flips). If you stretch that back all the way to the 1920s or beyond, then it gets really dicey in terms of discerning when a county actually flipped; you have to create new shades in between the existing ones.
From a pragmatic standpoint, I actually think 1964/1972 are the most ideal points on balance to show this, seeing as how they were the last two true landslide elections for each party. It gives you an idea of which counties are truly the most solid counties for each party in the quasi-modern era, as well as which ones weathered the storm even throughout those elections.
EDIT: Yes, it's possible that select years could be grouped together to avoid such an issue, as well as a range of colors. Aesthetically, though, it'd be less enjoyable I feel.