[halspero]

Problem is that knowledge can't be measured because it's not a list of contents that can be checked off. Deep knowledge in one area is by its nature integrated with other areas. Thus exams and testing make as much sense as ranking authors by spelling bee.

[throwaway729]

In some sense, this problem at least indicates we're trying to do something right.

Some knowledge is both necessary and testable: arithmetic, geometry, algorithmic solutions to common problems and properties of those algorithms, spelling of common words, basic grammar, many scientific facts/ideas/techniques.

The learning objectives of Common Core and other curriculum are ambitious -- they want to capture that "deep knowledge" you mention, and are often informed by input from subject matter experts.

However, the realities of modern schooling and assessment are not sufficiently resource-rich to fulfill and measure fulfillment of those standards.

[halspero]

Well physical quantities like length, time can be measured. Also fungible goods like wheat, oil. But knowledge is differentiated, non-homogenous and subject-specific. So it isn't measurable. Memorised facts like spellings, historical dates and common types of exam questions can be counted but these aren't useful in themselves. Knowledge is information that can do things.

[throwaway729]

I feel like knowledge is one of those "I know it with I see it" type things. It can be assessed, but not systematically via a multiple choice standardized test. Which puts a lot of trust in the judgement of teachers.

Not exactly a politically viable solution in America.

[halspero]

"I know it with I see it". Yes indeed. One has to interact with a person to know that he knows stuff.