>Evolution will have hard time identifying genes to promote, when the minimum unit of change is probably "how smart people are" or "how social people are", both having many more reasons for being favoured by natural selection.
If I understand you correctly, that’s just not true. In his seminal work, The Selfish Gene, Richard Dawkins showed that the individual gene is the unit of selection. A genotype that tends to result in a slightly stronger preference for a family life is exactly the kind of thing that could become more prominent, even within one or two generations. I’m sure this is happening right now.
>In your particular example, this arguably derails biological evolution completely - the decision to have kids, and how many kids to have, is dominated by economic and (secondarily) cultural considerations, which change significantly between generations.
I was not making a particular example at all. That is the function of natural selection: it selects for the genotypes had by those who have children who go on to have children themselves.
The selection doesn’t have to happen at the level of a preference for kids. A gene related to diligence or conscientiousness or just about anything else can have effects that ultimately cause differential reproductive success for you and your offspring.
The fact that it’s all too complicated for us to model has bearing on it at all.
Gould came up with the excellent name "punctuated equilibrium" for the discontinuous nature of genetic change, though the math describing it had existed before he popularized the concept.
It is true, though, that all the big changes of the last 10,000 years seem to have been simple changes in individual genes or their expression rather than more complex changes.
Here's a reality check: 6-finger people also have a much higher probability to be bullied at school and thus never reproduce than the others.