The Framingham study is a good study for what it is about (risk factors for heart disease), but the study is probably inadequate for studying general selection pressure on the whole human population.
Degenerate diseases don't kill you before you can have children.
"Those changes we predict for 2409 could be wiped out by something as simple as a new school-lunch program."
Stupid. African nations have both the highest birth rates and the worst nutrition. If anything, expected good health after reproductive age negatively correlates with fecundity.
"Uniquely in the living world, what makes humans what we are is in our minds, in our society, and not in our evolution."
If Steve Jones knows of some supernatural force that excludes the operation of the human body from physical reality, he should describe that force and submit it to a physics journal. Otherwise, yes, minds are minds, but minds are not magic: they're meat.
Norvig's advice is cautionary not contingent. When you want to study the effects of natural evolution on human populations (really any population) you're forced to do a wide, long observational study including as many subjects and cofactors as possible. Since that's pretty nearly prohibitively expensive, you're really better off just seeing if you can get someone else's data just like Stearns did.
This data specifically doesn't have a lot of power to generalize to populations outside of Framington. Additionally, making bold future predictions is not something that scientists like to show one another, just news reporters. What it does, though, is show support to already leading theories. Stearns is quoted in an NPR interview saying the real point that "the result that we have here is no surprise to any evolutionary biologist. It's exactly what we expected to see."
No matter what the possible confounding factors may have been, the fact that when decent statistical methods were used to control for and limit just to correlations that could be feasibly tied to evolutionary processes the remaining effect fits the predictions developed by theories made prior to this result means that the data provided some interesting justification.
Stearns also is quoted in that the study attempted to "lift the level of discussion and get everybody on the same page." It's not groundbreaking, it's just some solid human-based evidence for things long predicted. The hype is centered around news stories which pick up on the wide future predictions (an offhand, light comment in the interview) and the philosophical/moral implications of evolution playing police state on our futures (which is a suitable target for arguments about statistical weakness and Stearns himself brings up that only about 5% of the variability was involved in this evolution trend).
All the study really says is that, yes, like we thought, somewhere deep inside human cultural motion is a definite seed of evolutionary guidance. It takes 60 years of intense observation to see it, but it almost certainly exists and plays some tiny role.
http://norvig.com/experiment-design.html
The key quotations in the article are
Douglas Ewbank, "Those changes we predict for 2409 could be wiped out by something as simple as a new school-lunch program."
Steve Jones, "Uniquely in the living world, what makes humans what we are is in our minds, in our society, and not in our evolution."