I think it's time again for another issue of my signature hobbyist experiments in computational sociodynamics. This time I will try to shed some photons at the phenomenon of beauty (the 2026 edition!).
In the following I will, for the sake of simplicity, write strictly from my own perspective as a man. To apologize for your possible inconvenience, I want to cite one of my childhood heroes, the sailor Popeye: „I yam what I yam!“.
I have clear memories of my childhood crushes. I was madly in love with all of them. Surprisingly, none of them was a particularly „pretty“ girl in the sense that she had a very regular face. They were lively girls with lots of character, and this was also visible in their visual features. I was a rather shy boy when it came to girls. Ashamed of my feelings (more about this soon!), winning their heart was bound to remain a dream.
And, if we look back at pictures of movie stars from - let’s say - the 50’s, then we notice easily that many would barely meet modern beauty standards. And still they were the most desirable women of their time.
If I look at the female sex symbols of 2026 I have very mixed feelings. On one side I understand why other men find them attractive. Sadly there is, more than ever before, a shared standard regarding the „beauty“ of women among men now („she is a clear 9!“). But on the other hand I find these women also terribly boring. They somehow all look exactly the same. I cannot imagine spending two hours on a restaurant table with one of them, not to mention to start an erotic adventure [3].
What happened? What went wrong?
To understand this, a phenomenon observed in the animal kingdom might offer helpful clues. It’s called „fashion“ and works like this: in many species, individual animals tend to find those potential partners particularly attractive which are considered attractive by many other animals. This is a bit surprising, because to be considered attractive is not really an advantageous physical or cognitive trait which can support survival and procreation (like strength or intelligence). But this behavior is still far from pointless:
To determine the effective fitness of a potential partner is very difficult. But the perceived attractiveness (by the group) is real and sometimes possible to observe. It has, even if it is not connected to physical fitness, a real impact on reproductive success. Let’s say a male is considered attractive by many females. This means that his male offspring will probably be considered equally attractive (which will make it easier for them to find partners for mating).
Therefore, the shared perception of high attractiveness is a valid reason for a female to approach this male too.
Now there are factors limiting the impact of this effect. Under normal circumstances, attractiveness is not always easy to observe/measure. Evolution in humans has invented shame (introduced with me as an example above) to hide strategies from others. We instinctively hide our true feelings for potential partners from others to avoid making them needlessly more attractive for others: if I confess to my community to have a crush on a girl, other men could start to have a closer look at her too and my chances would get, due to increased competition, smaller.
But in the past decades we have changed something important: we have introduced the possibility to openly express our admiration for certain people without shame to our peer group: we can like fotos of men/women on social media. This is completely risk free for us, as everybody knows that we will never have the chance to get in contact with these stars and influencers. It's only a kind of platonic infatuation.
But, surprisingly these public beauty ratings have a strong impact on society nonetheless. They are easily visible for everybody (in the form of the number of followers or likes for fotos etc.).
To find out if (and how exactly) this could work out, I wrote a simple python script to simulate the impact of the "fashion" effect explained above in such an environment.
In the following, I try to give a non-technical explanation of how it works.
First I define a population of around 200 „people“. Each person gets a randomly assigned sex (women to men about 50:50) and random values for about 10 physical attributes. This could be things like body height, eye distance, nose length, shoe number etc.. But I don't really care about this, for me it's just a bunch of attributes. Values close to zero mean „average“ and negative or positive numbers mean that the attribute of this person is below or above average. The more all these numbers deviate from zero, the more personality the person has (this is a definition!)[2]. These numbers are fixed and will not change (we don't consider cosmetic surgery for now).
In the same way every person gets the same number of attributes assigned a second time. Again random values are used, but different ones. These numbers, the desired features, describe which kind of person (i.e with which physical attributes) the person finds most attractive. Unlike the physical features, these numbers can (and will!) change.
Then the script goes through every single person of the population. For each of them, first another person of the opposite sex (the so called target) ist chosen. Now we do several things:
- We calculate how attractive the person finds the target (i.e. the attractiveness). To do this we calculate how different the target's physical features are from the person's desired features. The smaller the difference, the higher the attractiveness.
- Now we check if this attractiveness (calculated in 1.) is above average or not. If it is, we update the persons desired features to be a little bit more like the physical features of the target. This shift is chosen to be proportional to the attractiveness of the target (note: this step 2. implements the fashion mechanism!).
- We update the targets average attractiveness with the new value from the person („klicking a few times on the like button“). Note: each person starts with the same value for the average attractiveness (zero).
As usual, you can find the script on my Github-Account.
This whole process is repeated several times („episodes“). Then we sort the population by their average attractiveness (which we updated in step 2. above. This corresponds roughly to the „number of likes/followers on social media“).
Now let's have a look at the output of the script:
Most / least attractive:
Attractiveness= 6.3708 Personality= 1.6798
Attractiveness= 5.4895 Personality= 1.8361
Attractiveness= 5.0612 Personality= 1.7459
Attractiveness= 5.0117 Personality= 1.8855
Attractiveness= 4.9427 Personality= 1.8976
...
Attractiveness= 2.6241 Personality= 3.4903
Attractiveness= 2.5744 Personality= 3.4184
Attractiveness= 2.5402 Personality= 3.7739
Attractiveness= 2.5185 Personality= 3.7527
Attractiveness= 2.3139 Personality= 3.9642
Those people with the most average features (i.e. lowest „personality“) end up as the most attractive!
Now you might object „this is nothing new! It's known for years that by averaging over many fotos of faces we can create new synthetic faces which are perceived as exceptionally beautiful“.
True! But - and this I find quite interesting - such a behavior rule does not appear anywhere in my code!
Therefore: this phenomenon, which we observe today more clearly than ever before, is only a mathematical artefact of the fashion mechanism (which got a bit out of control recently „thanks“ to social media)[1].
Our preference for the average, the boring is not millions of years old. No, this is new, we created it recently!
I will explore the mathematical reasons for this result in more detail in an upcoming post.
[1] Very well possible that this conclusion is also not new (I'm not an expert in the field). But I hope you find the text interesting nonetheless.
[2] We could also call this the person's ugliness. But I'm sure that, after reading this post, you'd find this very misleading too.
[3] I'm in perfect hands!
Image on top: Pixabay / Med Ahabchane
Sophia Loren making something like Pizza around 1965 (Good Lord!):
(Image: embedded from Getty Images with their fishy looking JavaScript snipplet. I hope it will not erase your harddisk. Sorry, but I did not want to pay CHF 150.- for a foto. Not even for her foto)