Yann Klimentidis points to a preprint of a paper by Eyre-Walker et al. coming out in Genetics entitled The distribution of fitness effects of new deleterious amino acid mutations in humans. The abstract finishes with this:

We argue that the relaxation of natural selection due to modern medicine and reduced variance in family size is not likely to lead to a rapid decline in genetic quality, but that it will be very difficult to locate most of the genes involved in complex disease

These are two possibly interesting points; let's take a closer look at both of them:

1. Decline in fitness due to medicine (I've never heard the argument about reduced variance in family size, and it's bizarrely never mentioned again in the article). This is a classic worry of geneticists, dating back to god-knows-when but popularized as the notion of "genetic load" -- the difference in fitness between the average individual in a population and some ideal. The argument here is that some mutations that would be deleterious in an "ancient environment" no longer have a fitness cost due to medical advances, and thus are free
to drift through the population. If by chance the "ancient environment" were to then re-assert itself, most people would have such a low fitness that it could be the end of the population. Of course, I sincerely doubt anyone stays up late at night worrying about this possibility.

But what people occasionally stop to think about is a more subtle argument: an increased genetic load means an increased dependance on technology to offset the potential fitness costs. An example: imagine there's a genetic disease that in the past lead to decreased fitness, but is currently treatable with a little pill. If the disease becomes more common (it's not being acted on by selection, after all), more and more people will need the little pill, and, in the long run, the entire population could theoretically be affected. Thus, efficient treatment of disease could paradoxically lead to more disease and more money being spent to treat it. This is somewhat more reasonable a fear, but still unjustafiably paranoid. Most populations are already dependant on technology to one degree or another (and imagine the train wrecks our genomes are because of fire or agriculture...those innovative bastards!), and the use of technology to temper the unpredictable or inconvenient effects of the environment could probably be considered a uniting theme in all of human history. And of course, genetic engineering could (in theory, at least) render this point moot in the long run.

But anyways...Eyre-Walker et al. take their estimate of the number of deleterious mutations per person per generation (1.8) and their estimate of the average effect of a deleterious mutation (a loss of fitness of 4.4%), and come up with a maximum "decline in genetic quality" due to relaxation of selective constraints of 1.8*4.4= 7.7% (compared to some ideal population). And given that complete relaxation of selective contstraints is implausible and they think they've overestimated the average effect of a deleterious mutation, they decide the "decline in genetic quality" is likely to be more on the order of a couple percent per generation. Eh. Take from that what you will.

One last thing that essentially sums up why I find this whole thing a bit silly: the subheading they write this all under is "Decline in fitness". But if you define fitness in any classical manner (reproductive success, for example), medicine doesn't lead to a decline in fitness at all, but often an increase. If they wanted to be precise, the subheading would be "Hypothetical decline in fitness were medicine to disappear overnight" or "Decline in genetic quality if we consider 'quality' a measure of adaptation to an ancestral environment". When you think about it like that, it's a lot less ominous. Imagine the decline in fitness if we went back to being hunter-gatherers. Modern medicine, like agriculture, is now part of our environment (at least where I live), and our genomes will slowly adjust to that fact. Nothing to worry about there.

(This has gone on a little long, so I'll address the claims about
finding disease genes in a separate post)