It's possible, however, that Kristof is just "cherry-picking problems from a test on which American students, on average, do pretty well in comparison," according to the article.
Essentially, Kristof uses the Trends in International Mathematics and Science Survey, a standardized test of eighth-graders, but only uses the questions that students did worse on than any other.
As a whole, the Americans didn't do that bad on the entire test -- in fact, for the majority of the test, they did better than average. Overall, American students scored slightly above average, on par with Finland’s test-takers.