Hacker News Reader

Comment by nine_k | Hacker News Reader

If exam grades follow some absolute standard, then throwing better students at them would result at more As, up tp only A grades for an entire class. This is without any grade inflation: the absolute level of knowledge to get an A stays the same, but the absolute level of knowledge demonstrated by students goes up.

To the contrary, to have any B or C grades in a class full of the brightest students, one needs to either increase the absolute expected level, or to give the class grades relative to its strength, which again means raising the expected level for an A.

Exam questions from Harvard can be easily tracked 100+ years back. A researcher who'd give current Harvard graduates some of the older questions could find out if the toughness of the exams went up, down, or stayed.

View on HN

> then throwing better students at them would result at more As

Also, the Flynn effect: "the substantial and long-sustained increase in both fluid and crystallized intelligence test scores that were measured in many parts of the world over the 20th century" - https://en.wikipedia.org/wiki/Flynn_effect

> A researcher who'd give current Harvard graduates some of the older questions could find out if the toughness of the exams went up, down, or stayed.

I strongly suspect this wouldn't be informative. The base knowledge of what we expect students to know has shifted.

Some might be topical: "How did Harding conduct his campaign?" "Name the primary components of a steam locomotive."

100 years ago I believe there was still an expectation that students knew Latin. The 1869 Harvard entrance exam at https://graphics8.nytimes.com/packages/pdf/education/harvard... has two pages of Latin and two of Greek.

(Most of the student then would have come from an upper-class prep school, where they would have a training that let them be able to answer "Bound the basin of the Po, of the Mississippi, of the St. Lawrence." and "Supply the two names left blank in the following passage from the Oration for the Manilian Law.")

Consider how calculus is now regularly taught in high school, while pre-war high school ended with solid geometry.

Or, consider how an English literature course 100 years ago (and its tests) would have focused on the literature of England and the US, while the equivalent modern English literature course would likely have drawn from the much larger Anglo-sphere.

How would you determine if the toughness of an exam from a hundred years ago matched a modern day one? The pedagogy, tools, and knowledge base differed dramatically in the past century. A question might look difficult but be intended as a take-home, be graded differently than binary right-wrong, rely on a trick of an outdated tool (ex. slide rule).

> Exam questions from Harvard can be easily tracked 100+ years back. A researcher who'd give current Harvard graduates some of the older questions could find out if the toughness of the exams went up, down, or stayed.

Agreed!

This would be great! My guess would be in 2022 the math is harder but literature easier.

It would need to be done both ways.

A 1923 student is not going to have read "The Gulag Archipelago" or "Things Fall Apart" or other works published after 1923, so would fail any modern questions which require having read those works.

But even if you look to contemporaneous 1923 literature, I strongly suspect a modern Harvard student is more likely to have read black literature from 100+ years ago than a 1923 student. (For examples see https://histlit.fas.harvard.edu/files/histlit/files/america_... with "Sample Oral Exam Questions" from a modern Harvard exam section on "Being Black After Reconstruction".)

If the 1923 student didn't have to read 75% of what modern literature courses cover, and instead read a different 75%, perhaps with a stronger emphasis on British literature, then of course a 2023 student won't test so well on a 1923 test. But then, a 1923 student wouldn't test so well on a 2023 test.

Even math has some issues. Here's a Harvard entrance exam from 1869: https://graphics8.nytimes.com/packages/pdf/education/harvard...

You can see the math part requires details that aren't taught now:

"Find the amount of £50 12s. 5d. at simple interest at 8 per cent., at the end of 5 years 2 months and 3 days."

"From 1 sq. rod 5 sq. ft. subtract 7 sq yd. 139 sq. in."

The Brits decimalized a half century ago, so kids aren't taught there are 20 shillings in a pound and 12 pennies in a shilling. Only surveyors use rods these days, and it's defined as 16 1/2 US survey feet - which is slightly smaller than the international foot defined in 1959.

Or, "Find the cube root of 0.0093". By hand. We no longer consider that something worthwhile to teach.

They aren't hard, but without that small bit of knowledge or training, modern students are very unlikely to get the right answer.

Or, "State and prove the proportion which exists between the parts of two chords which cut each other in a circle. State what proportion exists when two secants are drawn from a point without the circle." I don't know about you, but my plane geometry isn't good enough to answer that question.

That's due to post-war educational policy changes that de-emphasized classical geometry in favor of calculus.

In the History and Geography section of the 1869 test I'm struck by the emphasis on rivers. Rivers are still important for transportation of commodities, but not nearly as much. I can't see them being emphasized in a college entrance exam today. I might be able to bound the basin of the Mississippi, and state the source of the Amazon and Ganges (if I guess correctly, and have a labeled map). The rest, no chance.

Geography seems to have been a more important topic then. The Sandhurst exam, 1890-ish, included "draw from memory a map of some country or other", quoting Churchill from https://archive.org/details/in.ernet.dli.2015.76347/page/n27... :

> I also succeeded in passing the preliminary examination for the Army while still almost at the bottom of the school. This examination seemed to have called forth a very special effort on my part, for many boys far above me in the school failed in it. I also had a piece of good luck. We knew that among other questions we should be asked to draw from memory a map of some country or other. The night before by way of final preparation I put the names of all the maps in the atlas into a hat and drew out New Zealand. I applied my good memory to the geography of that Dominion. Sure enough the first question in the paper was: ‘Draw a map of New Zealand.’

I can well imagine that a military officer may need to have that knowledge at hand faster than consulting a map. ("I am the very model of a modern major general!" and all.) For upper-class U.S. Harvard students, trade knowledge may be more important, thought that doesn't explain needing to know the source of the Amazon.

In any day and age, education for the elites is only in part about what's actually useful. The other part is there to segregate itself from the hoi polloi - and it actually makes more sense for it to be something not immediately useful precisely so that it's a clear marker of status: "I have time and money to learn this, of all things".

Good point, thanks!

Though of course it needs to be the right "this".

People often can't answer these question today because the answers are now misleading or irrelevant, except to historians.

Even math has advanced in the last 100 years. Students today would be taught pertinent math, not how to use a slide rule. Throw a question asking for manipulations of a slide rule and 98% of today's students aren't going to be able to answer it. And most of those few who can are going to have a crotchety grandparent. The remaining few will be mathematical historians.

A while ago someone mentioned an old test from an American high school somewhere. One of the questions on that test was "name the organ responsible for purification of the blood".

This question is impossible to answer because there is no organ uniquely responsible for that. It's an important function served by several organs. Most prominently:

1. The kidneys are responsible for filtering certain toxins out of the bloodstream and ultimately getting them excreted in urine.

2. The liver is responsible for filtering other toxins out of the bloodstream and ultimately getting them excreted in urine.

3. The spleen is responsible for filtering defective blood out of the bloodstream and ultimately getting it excreted in feces. (Which is why feces is brown, incidentally.)

Without having attended a 19th-century anatomy class, there's no way for me to know what the person who wrote that test was hoping I would say. It seems pretty likely that I know more than he did, but that won't get me a good score on his test. Was that an easy question? Hard?

Engineers are taught how to use a slide rule. Mathematicians don’t need one!

(Just look at the engineering / math departments of today. Engineers make heavy use of calculators, spreadsheets, Matlab, etc. Math majors don’t.)

Don't math majors use Mathematica, Coq, or other such symbolic computation tools instead?

Not if you want peer acceptance and not be dismissed as a mere CS student.

It's acceptable to download your brilliance into something such as, say, CAYLEY/MAGMA [1], [2] but, obviously, once you start grinding the organ [3] and using it algorithmically for computation you're just another monkey . . .

[1] http://magma.maths.usyd.edu.au/conferences/london93.html

[2] https://link.springer.com/chapter/10.1007/978-3-642-15582-6_...

[3] http://magma.maths.usyd.edu.au/magma/

Tools don't often help you understand maths, but obviously help you to solve problems.

There are plenty of areas where tooling is rudimentary or unfashionable, like programming, where the tools are like chisels and saws used by artisans.

Reply on news.ycombinator.com