Ok, I'll try to explain it more thoroughly.
When you create a system of preferences for your dwarves,
you make it very roughly dependent on some kind of personality factor that interacts with the world.
Let's say dwarf loves tradition, then this dwarf probably likes some features that belong to some historical king or queen.
Dwarf is very family oriented - then he/she takes some preference traits from parents.
Now different possibilities will arise - in very traditionally minded culture, this might create more strict aesthetical standards, making things a little easier for those who bear these standards. However when you take more different personality traits into account, like family as in my example, these preferences might totally nullify eachother which in case make no general trends in macro-social scale. Or they might amplify eachother or they might even fluctuate - because world goes on and different events shape different social dynamics. Overall cultural values of civilization might also change (which is mechanic that is not implemented yet) when something drastic happens.
Your position is that sexual selection creates some *already* universal traits which is technically correct of course. However, only when you don't take larger historical events into account. Lets make an example and say 95% of civilization members get suddenly killed. What happens then? This type of bottleneck can make rare traits becoming universal and vice versa.
You have got to remember what we are trying to accomplish here. If we only have one partner per lifetime we have no need to go into elaborate mechanics, since a random system will work perfectly. The only reason to give a dwarf a definite preference/aversion for particular traits is if there are more than one partners possible for a dwarf in a lifetime, that means that we could compare the partners of said dwarf to eachother to discern what it is that they like.
If you make a system highly complex, as you are doing then you might as well make the system random. There is no need to go into the exact reasons for every single preference, because there are just too many factors involved for them to all be taken into account and they are either going to result in an essentially random result or we get the sexually selected uniformity we discussed before, hence making the whole system redundant.
A bottleneck does not get you anywhere here. If the traits are universal then a bottleneck will do not nothing to change that, while if traits are diverse it will simply reduce the diversity that exists. Since any selective system with non-random outcomes results in uniformity, a bottleneck simply makes a completely uniform system, more uniform.
How? By using thought police? Peasants are pragmatic breeders.
Because normative standards of everything are established by those with the power to do so. You are right of course in that most people simply ignore the normative standards, but that is not my problem, since my claim is that things are pretty much entirely random anyway because if they are not random then sexual selection would have already made the trait universal.
But this attraction to social cues (dress style, beard style etc) clearly is *not* universally fixed to the rulers dictates - Fashion has changed constantly (if at varying rates) throughout the entirety of human history - your theory is disproved by googling “16th century fashion” and then breaking down in a fit of laughter.
Of note is however that attraction to attributes is often caused by single or small groups of individuals who possess that trait (or outlandish fashion) in addition to others that are prized (Star athlete, dashing hero adventurer, heir to the throne), causing courtly imitation that then trickles down to the populace
It is not my theory, it is VislarRn's theory that I simply accepted for the sake of argument to point out how it did not work as a counter-argument to what I originally wrote.