One reason for its imperfection is that rather than measure problem solving or deductive reasoning abilities, it tests for knowledge previously attained. Unlike an IQ test, Wordsum performance can be significantly improved by preparation, even if the specific words included in the test are unknown by the test taker ahead of time. According to my college psych 101 course, this demonstrates the differences between assessing a test-taker's fluid intelligence (which IQ tests mostly do) and crystallized intelligence (Wordsum). The two are highly correlated, however, which is why Wordsum results provide a useful approximation of IQ scores at the group level.
Crystallized intelligence is said to increase with time, as the accumulation of knowledge and experience builds. But at some point, the destructive forces of aging set in and begin attacking crystallized intelligence, the assault on fluid intelligence having been well under way for several decades.
By looking at average Wordsum scores by age range, the GSS allows one angle from which to look at the decline in crystallized intelligence and when it tends to begin. The following graph shows as much. To avoid racial confounding, non-white scores are excluded. For contemporary relevance, all results are from 2000 onward (n = 4,072):
Some noise notwithstanding, there is a steady increase in mean Wordsum scores from the late teens into the mid-sixties, at which point decline sets in. People tend to enter retirement when they're minds are filled to the brim. Those who are forced to work into their retirement years are not afforded the luxury of going out on top. It must be depressing to experience a seepage of knowledge in one's chosen career after potentially having spent an entire professional lifetime accumulating it.
How nice it would be if we were able to reverse the declines aging inevitably (or evitably, perhaps?) brings.