In Phil's post about Children and Power, he describes a typical worksheet and the kind of mental trickery that these exercises often embody. The exercise he describes is a type example of tedious busy work: it's "make work" for students constructed in a particular style to make it easy to correct.
The biggest problem to good assessment is our society's assumptions that answers can be "right or wrong". This simpleminded attitude has nothing to do with learning or judgement and everything to do with making tasks that are cheap and simple to evaluate. There are no questions that have simple answers: even "What's 1+1?" can lead to a whole discussion about the nature of integers or the literary origin of using 1+1=3 to talk about emergence (the sum being more than the whole of its parts) -- and that's just scratching the surface.
The most evil outcome of this system, is giving machines the task of evaluating human productions. For meaningful learning, human productions need to be evaluated by humans that can appropriate a statement (ie, put into the context of the larger conversation) and then help the student see how their production fits -- or does not fit -- what the teacher had in mind. There was a school of thought in cybernetics called "Programmed Instruction" that tried to create systems that could do this, but it runs into the fact that domains cannot be fully specified. Programmed instruction went out with behaviorism, although you still see people every few years, ignorant of the history, who assume it should be easy to do.
I did some freelance work for a text book one time and one of the things they wanted desperately was for everything to have some kind of "assessment" associated with it. Every chapter, every section needed to have assessments. When they ran out of space in the book, and wanted to add these random grab-bags of facts online, they even wanted those to have assessments. And that was when I made the realization: when I pointed out that it was meaningless to have assessments of random collections of facts, which had no underlying conceptual dimension, they just hired someone who was desperate enough for the money to write a bunch of questions. That's the lowest common denominator here: these assessments get designed as cheaply as possible, no matter how meaningless and pointless they are.
For students to be engaged in tasks, the tasks need to have some purpose. They need to be things the students think are worth doing and part of a larger effort that's going somewhere interesting. Schoolwork that is pointless and a waste of time encourages students to be cynical about the whole enterprise and encourages cheating.
It's possible to have students to real work -- work that matters -- that requires the same kinds of skills. And there's no reason not to do this, except that it requires three things. It requires teachers have (1) the freedom to let students go in different directions, (2) the wherewithal in terms of time and imagination to not just grab the first worksheet that comes to hand and co-construct interesting tasks with students, and (3) the time to provide meaningful mentorship and evaluation to students as their projects develop. Unfortunately, as budgets are cut and teaching profession becomes increasingly deprofessionalized, none of these are likely to happen in public education anytime soon.