Hacker News Reader

Comment by jellicle | Hacker News Reader

> One approach to answering questions like this is to build a little model

This test question isn't testing that.

> What I’m saying is that answering this question well requires experience with microscopes or similar devices.

It sure doesn't. It requires you to be able to read and comprehend text. That's it. The subject could have been Martian fizzbozzles instead of microscopes.

> Being able to answer the question without prior knowledge of what a microscope is, or how light behaves, or lenses period, or refracting telescopes, and so forth, is ridiculously hard.

It sure isn't. Anyone who can read and comprehend text can answer it with no prior knowledge of the subject.

View on HN

My HN friend, my argument was that the writer isn’t daft, and neither is his complaint, and I stand by that. My argument is that the writer raises interesting and useful questions, and I stand by that too.

As to the fact that the question doesn’t test internal modelling, of course it does. If you made it a question about something nobody has ever seen before, then it wouldn’t test knowledge of microscopes. But it does ask about microscopes.

Perhaps, in aggregate, the test covers enough subject areas that the students tested will be exposed to a roughly equal mix of things they have experience with and things they don’t. But as I noted elsewhere, the fact that it’s possible to answer this question from the text alone in no way changes the fact that it’s easier to answer the question with prior experience. And from that, I think it’s perfectly reasonable to ask whether this is a good question.

Yes, if it were truly just about reading comprehension, then the test-writers would be horrifically incompetent, and therefore unfit to test anyone. Anyone who competently designs tests (experimenters, pollsters, etc) knows to address such biasing factors.

But I think the testers aren't total morons, and know that such a question is biased towards children with proper (privileged) backgrounds where microscopes and the respective training are commonplace. They'll be faster at answering the question and not feel ignorant — crucial advantages on tests.

(The testers may not even like this facet of their jobs, like ad industry developers who use ad-blockers, but it sure beats unemployment.)

And the whole question smacks of "scientism", where tools are confused with science. You never see a question like, "Scientists have believed something for 200 years. Is it necessarily true?"

> It requires you to be able to read and comprehend text. That's it. The subject could have been Martian fizzbozzles instead of microscopes.

And therein lies a great fallacy which is still widely believed in our post-Enlightenment world. But the map is not the terrain. A mental model cannot be completely represented with words or pictures, or anything. Words and pictures, even good ones chosen for pedagogical value, are only pale projections of a complex thing onto a flat surface.

>Anyone who can read and comprehend text can answer it with no prior knowledge of the subject.

Without qualification, this statement is strictly false. There may be certain kinds of text that can be comprehended with no prior knowledge of the subject if you assume a certain baseline of internalized knowledge. There are many old parlor games (which have sadly gone out of style) which rely on hacking this qualification. For example, one of them requires that one person follow the instructions of another. One person tries to describe how to draw a particular geometric figure. The person doing the drawing must follow the instruction, but they also are free (in fact encouraged) to ignore the intention of the instruction. It's a game that puts the difference between words and reality up-front and center.

You put this wonderfully. People are very little more than our experience with the world itself. Without external stimuli, we would be trapped within our own minds and only be capable of evaluating a limited set of possibilities. Merely the act of touching an object generates complex cognitive experiences. This is not possible by abstraction alone.

I understand what you're getting at, but that's not the case in this question, since one of the four answers are stated almost word for word in the text. (caution point 4)

The rest of the alternatives are "plausible guesses" which you might think are reasonable if you don't remember the point above. So what they're REALLY testing is if you read the text carefully enough to remember that line or not.

Actually having knowledge of microscopes are probably a bad thing on this particular question since you're likely to skip over the boring warning part, like I did when I quickly read the question just now.

Did you read what raganwald wrote? He's saying that reading comprehension is not typically about memorization, it's often about model building. As a human reads they don't just fill up big bin with data, they build up models. When reading a narrative we will build up ideas of characters. We'll fill in more and more of their details. And when probed about the story we will fall back to our knowledge of those characters, and it may take us a while, if we even can, to figure out exactly which events in the narrative contributed to each aspect of a character.

In this case the object of the narrative is a microscope. And as people read through the text they aren't memorizing sentences or facts they are filling out the model, or character, of the microscope. But people who already have an accurate model of a microscope from personal experience will have an easier job doing so. It's similar to the way standardized tests can be racially biased. If I put forth a "story problem" about the rules of football, being familiar with those rules already gives one an advantage.

> Anyone who can read and comprehend text can answer it with no prior knowledge of the subject.

That's the writer's complaint. In order for this question to be testing reading comprehension rather than laboratory experience, it must be assumed that the test taker has not used a microscope. Otherwise, they could answer the question without even looking at the text.

> It sure isn't. Anyone who can read and comprehend text can answer it with no prior knowledge of the subject.

This reminds me of people who think the manpages are a sufficient resource to completely understand the unix cli environment.

Reply on news.ycombinator.com