I’ve recently hit upon yet another improved face-averaging process, and to try it out I’ve turned (among other things) to one of the first datasets I experimented with back in 2014: portraits of the candidates in the Miss America competition. I hope you’ll find the results interesting. But the institution of Miss America itself has evolved since 2014 in ways that arguably complicate my project.
The year 2019 witnessed the launch of “Miss America 2.0,” in which candidates are no longer to be judged based on outward appearance. Until 2018, I take it, they were. Now it’s all officially about empowerment, eloquence, leadership, and talent, which is surely a positive step forward into the twenty-first century. For my purposes, though, the main consequence is that the official portraits posted on the Miss America website went—in what I take to be a symbolic gesture—from color to black and white, with a far greater diversity of poses and expressions besides. Here, then, is an average of the official portraits of candidates for the year 2020:
And here’s the same for 2019, the first year of Miss America 2.0:
I can’t fault Miss America 2.0 for introducing a new style of portrait to mirror the changes in the format and emphasis of the competition. On the other hand, it’s hard to find groups of stylistically similar portraits spanning long periods for image-averaging projects such as mine, so I have some mixed feelings about the discontinuity. With that in mind, I was happy to find an alternative set of portraits in color for the 2019 candidates, published in August 2018, which are a lot more like the official portraits of the past couple decades. I suspect these might have been intended to function as the official portraits until a decision was made to go monochrome instead. Nothing comparable for 2020 seems to exist.
At first glance, the move of Miss America 2.0 away from outward appearance might seem to call into question the relevance of face-averaging as a source of insight into the essence of each year’s portraits. But in some ways the results could be even more informative than before. The facial expression and orientation seen in the average of color portraits for 2019 is basically the same one we get from averaging similar sets of portraits dating back a number of years. It can be regarded as the conventional pose of Miss America one-point-whatever. By contrast, the average of the black-and-white portraits for the same year reveals a more serious expression, an upwardly raised face, and only the faintest hint of a smile. The average of portraits for 2020 then shows the face tilted back downward, with a broad smile once again—but one that’s now more exuberant than polite. It’s improbable that the candidates for each year just happened to fall into these patterns in pursuit of their own self-expression. Presumably a photographer instructed them in how they should pose, within certain limits. The differences we see—and can see more clearly in the averages than in the individual portraits—would thus appear to expose different approaches to self-presentation which the Miss America organization has been experimenting with while seeking a path forward in a new cultural milieu.
By the same token, that last image might end up being the last one it’s possible to create as part of a long unbroken sequence of images in approximately the same (old) style. For each regnal year from 1997 through 2019 (corresponding to state competitions from 1996 through 2018), I’ve managed to secure color portraits of all candidates online—mostly official portraits that appeared at various times on the Miss America website itself, and almost always as complete annual sets; only rarely have I had to search out an individual missing portrait. At the end of this post, I’ll share annual averages for every year back to 1997, similar to the ones shown above. I’ll probably also have more to say in future posts about the averaging process itself, which is a marked technical improvement over what I was doing previously.
But I’d like to begin with a complete set of averages by state, which I think do a particularly compelling job of putting the averaging process through its paces.
Each of the state averages below is based on twenty-five portraits, but these represent only twenty-four annual candidates, since each candidate for 2019 is doubly represented by both her color portrait and her official black-and-white portrait. Not included are Miss Puerto Rico (competed 2011-2017) and Miss Virgin Islands (competed 2005-2015), for which there were too few candidates to create solid averages.
Many of the state averages look much like one another, but others came out rather distinctive-looking. Given that we’re dealing with averages across twenty-four years, these particularities are probably statistically significant—but I’ll leave it to you, dear reader, to decide what to make of them.
Many sites would force you to click through images like these one by one, each attended by a cluster of off-putting commercial messages (“Gut doctor tells you to get rid of this vegetable NOW!”). No such shenanigans for me. But if you find these images thought-provoking, kindly spread the word.
Now for the remainder of the year-by-year averages. Unlike the state averages, some of which are facially distinctive, these annual averages look as though they could all be pictures of the same person. But there are still at least subtle differences between the averages for adjacent years, and broad changes in fashion are easy to notice over the long term.