Haven't read the whole article, but my feeling is that basing eligibility to vote on passing a test might be a bad idea. It could end up meaning "you are allowed to vote, if you will vote for the right thing", where the right thing would presumably be voting for the people in power at the time of the introduction of the test.
The argument against that is that it has the same effect as the poll tax--keeping the lower classes out. It is claimed that this discriminates against the races with lower literate populations.
That's one argument, but I think the parent comment was making a different argument: whoever gets to write the test gets to decide what matters, and what guarantee do you have that the test will really be testing value-neutral relevant facts, or (even if it is) that the choice of which facts won't introduce bias? I expect it wouldn't be hard to come up with a test for which (1) almost all the experts agree on what all the answers are, but (2) if you select from the general population on the basis of what those answers are you get a sample that's politically biased. (Ludicrous example: suppose the test gives you lots of quotations from Ayn Rand's fiction and asks you which character said what. The answers are totally uncontroversial, but if only people who answered correctly could vote you might just happen to get a distinctly more libertarian government. Do something similar with the works of Karl Marx and you'd get a distinctly more socialist government. And so on. Of course in practice someone seeking to bias the test would do it in more subtle ways.)
