42 years from now...aren't such projections useless?
Not really. Projecting large populations is fairly easy to do accurately. We know the current demographics, and so we can extrapolate health care trends and throw an immigration guess in to find the upper half (i.e. people over 42) of our population pyramid in 2050. The lower half we can do slightly less accurately, as it requires an additional guess: birthrates (which is in turn sensitive to the immigration guess). The whole pyramid can probably be done to within a 5% overall error, barring a nuclear weapon strike or cure for cancer.