Hacker News Reader

Comment by TravisHusky | Hacker News Reader

Just to throw it out there, I graduated during the pandemic, and it was so frustrating not only the amount of cheating that occurred, which meant that if you didn't cheat you fell behind; but also how professors responded. I had a few professors who made exams ridiculously hard with strict time limits, just because they expected people to cheat. So the exams were designed in a way they cheating was the only way to get through it, which just encouraged more people to cheat. Not to say every person cheated, or every professor designed tests to be harder, but both groups were larger than you'd expect.

The good news is I think that with classes being in person again things will go relatively back to normal. Honestly I don't think I can fault students who cheated during the pandemic, it was just so abnormal that most people did whatever they had to to cope.

View on HN

I had an experience like this in Numerical Methods in university years ago.

The exams were incredibly difficult for bad reasons: you had to calculate many iterations of an algorithm by hand. You also had to memorize a bunch of algorithms (rather than understanding how they worked or how to modify them to solve different problems).

The professor also had a "can't do anything about it" attitude, and so the entire class was blatantly cheating as a result, and I got the worst grade of my academic career.

It sort of broke the camel's back for me; I consider our current approach to academics to be fundamentally broken.

It's harder to design exams where people cannot cheat, but it's certainly possible. Randomize the orders of questions on the test. Have multiple versions the same question (e.g. in a math exam, change a 2 to a 4, or a sin to a cos). Now you have unique exams for each student.

Most importantly, design questions that require critical thinking, and allow everyone to use their computers/books/notes/whatever. That's far more representative of real life anyway.

These simple changes would both make cheating very difficult, and result in better learning: memorizing a bunch of crap the day before an exam will have basically no impact in my life. Learning how to reason critically about the subject matter will give me lasting benefits, even if that subject never comes up again.

This ProctorU stuff is just quadrupling down on a fundamentally broken approach and making it much worse.

My AP calculus teacher allowed us to use calculators on our tests. Why? Because he knew exactly how to design questions such that the calculator would switch from exact to approximation mode. So you could enter any problem on the test into the calculator, and all it would spit out is a useless answer with as many decimal places as it could handle. Of course the calculator worked fine on every homework question. But you better have still learned the material if you wanted to make it through the test.

I don't follow. Do you mean to say that the calculator would produce .333333333 repeating but the student was expected to write 1/3 ?

I suspect think more along the lines of spitting out a numerical answer instead of sqrt(pi) when fed with \int_-\infty^\infty exp(-x^2) dx...

Those reciprocals are too small.

Lets see if the kids know how to multiply fractions:

factorial(256) times ( 1 divided by factorial(256) ) equals ?

And yes, there will unfortunately be kids trying to brute force it by typing it in instead of cancelling the factorial(256).

The TI-89 comes with a excellent symbolic solver that can do solve this or any calc problem and give the proper symbolic result. It can even do things like equation simplification and give the results in terms of the variables. Looks like this https://imgur.com/a/FmdCyYU

Saying any calc problem is a stretch. Just pull up the MIT integration bee and see how many the 89 can get symbolic results for.

I think the 89 is awesome for algebra because typing in an equation and doing an ans(1) with an operation will let you do the algebra process without making stupid arithmetic mistakes. I wish someone made a calculator that does this only and not one that does all the other CAS tricks for integration and derivatives. Oh and I wish they would make it RPN but that is a whole other story.

Yes, except not obvious and usually involving irrationals, so good luck figuring it out.

Calculators are generally useless in Calculus. Or at least they should be. Calculus is a process, and normally you are graded on that process, not on the answer.

We were allowed calculators too, but the problems were designed that anyone even mildly good at mental math didn't need them. In theory you could teach all of calculus without ever even using numbers (perhaps except 0).

It's been a while since I took calculus, but I seem to remember that a TI-89 calculator could show its steps when solving a problem.

Yeah I also had a TI-89 and you could directly type in most calc problem and get a great step by step answer. The symbolic solver on that thing was great.

Yep. We all had either TI-89 or -92. I used it to check myself after doing practice problems.

I think you need more numbers than that. At least 1, pi, and e. Definitely don’t need any numbers bigger than 4.

>Calculators are generally useless in Calculus. Or at least they should be. Calculus is a process, and normally you are graded on that process, not on the answer.

Yes.

It's been so long since I took calculus but that's the reason why we weren't allowed calculators in my Calculus class. All the formulas needed for the test were also provided (as we weren't being graded on our memorization skills either)

My professor also said don't bother doing simple arithmetic just to simplify an answer - leave it at 12/56 or whatever.

Turns out, it was a really, really great system.

> Calculators are generally useless in Calculus. Or at least they should be.

Mathematica, on the other hand...

Wolfram Alpha and Mathematica got me through Calculus. I paid for an account because I absolutely loved the "show step by step process" feature so you could actually learn how and why you were getting various results.

In calculus a calculator can often give you the answer but you only get credit for showing the steps which the calculator can't help you with.

Just get or a program a calculator that shows steps then.

I think professors should ask themselves if their exam would make a good CAPTCHA. If a machine could solve the exam easier than a human, it's a bad exam.

"circle all pictures that represent sqrt(pi)"

Calculus captchas do exist, which I guess proves the point!

https://nakedsecurity.sophos.com/2011/03/09/a-complicated-ca...

Have you considered that memorization is a good thing?

I memorized a lot of information, but I still search just as often on google and stackoverflow whenever I needed to check something.

It also help that I don't try to memorize a bunch of a crap before an exam.

You must not have seen a modern home cheating rig. The student read/scrolls through the entire test to read each question, as is recommended. A small camera that sees the screen transmits the questions to a team in another room. Each team member works on one question at a time. A single person then communicates the prepared answers back to the test taker via an earpiece, one small enough that it isn't detectable on the camera. Such a rig costs a few hundred dollars at most.

Another common trick: looking at your keyboard isn't cheating. Any small display secreted on your keyboard will go unnoticed. You will just look like you aren't a great typist, someone who needs to look at the keys.

Want to go real hardcore? Cut into your laptop and send the screen/keyboard/mouse signals to a different monitor. Then you can have an actor sit in front of the webcam pretending to take the test. I doubt they are correlating keypress sounds to characters appearing on the screen... yet. As any security expert would tell you: if the attacker has unfettered access to your hardware then all bets are off.

How much does the team cost?

This is one of those situations in which it pays to have likeminded friends.

That would depend on the level of expertise you need. To pass a highschool calc exam, probably less than 100$. To have someone sit the LSAT for you in person, probably in the low five figures. I've never been involved with people actually being paid for such things but they do happen. Huffman allegedly paid $15,000 to get her daughter unlimited time on the SAT. Paying to have someone actually stand in would have certainly cost more.

The team could just be your own friends in the class. Each one of you gets a turn to take the test with the others helping find answers. No cost.

Fundamentally, what you are complaining about is teaching being under resourced: making lots of variations of questions obviously takes more effort both to create and to score so that's why it's not done. Making questions that require critical thinking is very hard because you need a very good understanding of the students mind to test the correct thing, so it's usually not done.

It's also that many teachers are underproficient.

open book exams with serious thinking questions is the answer.

I had such a numerical methods exam, I had a python script ready for it. No regrets on automating something that should be automated. I understood how the method works and wrote code for it.

> I understood how the method works and wrote code for it

We were allowed (and encouraged) to write cheatsheets for most of our exams at university. However, they had to be a single sheet and handwritten.

The reasoning was that in the process of creating a cheatsheet, students would reinforce their learning. The handwriting requirement was to encourage us to write them ourselves, instead of just printing one off, and the size requirement was so that we had to think about how to distill a semester of learning instead of just trying to copy the entire textbook by hand.

I really liked the system, and I feel like it did help more than just allowing open book exams. I remember of the few exams that were open book (some even allowed us to use the internet, but with a ban on communicating with others), I usually got a worse grade, as I ended up putting less effort into studying.

We had a particularly articulate young man in our graduating class that argued that checking the book is ultimately not cheating in the world of professional CS and software development. Rather, it’s considered being responsible. So why should we be taking tests that treat it as anomalous or cheating?

He convinced one professor, possibly two, but if he had any more luck than that, I wasn’t in those classes with him.

Most of what we do is synthesizing facts into knowledge. But I know from copious personal experience that there is a wide grey area between recalling a fact at will and forgetting it exists entirely. I am fairly good at recalling concepts in general terms. Y might be applicable in this case, but I need to check the details to be sure/explain how.

Which is also why I think banning computers from planning meetings is stupid. You want us to agree to do things without being able to spot check any of our hunches that will drastically effect our estimates? That’s a winning plan /s.

Totally agree. Generally my response to that situation is usually, "wait you did that from memory without looking it up? That's super irresponsible, check your work or use an existing ___".

> We had a particularly articulate young man in our graduating class that argued that checking the book is ultimately not cheating in the world of professional CS and software development.

My school's CS department defaulted to allowing students to use almost any available resource (unless specifically indicated otherwise). But you were required to write down what resources you used.

So a professor might assign a whole pile of nasty algorithm proofs, and say, "You're allowed to work in groups to find the general solution. But you need to write down your group members, and you need to do your own writeups." Tests were in person but not especially proctored. The penalty for cheating was a heavy suspension and a transcript annotation, with permanent expulsion for a second offense.

This actually worked pretty well, back in the day.

> I had a few professors who made exams ridiculously hard with strict time limits, just because they expected people to cheat. So the exams were designed in a way they cheating was the only way to get through it, which just encouraged more people to cheat.

That’s such a strange response to that expectation. For the class I teach, I made my exams open book, open notes, open internet and gave them a extended time to work on the exam. If they’re gonna cheat, they’re gonna cheat. Don’t fight the tide. But the questions I asked where analysis/synthesis questions that couldn’t simply be looked up, and student performance was on a bell curve with a 80 average (in line with previous years). I think a lot of professors need to embrace the situation and adapt rather than forcing a police state on students during exam times.

I had a couple professors do exactly what you did and it was much more enjoyable. I feel like I leaned a lot more in classes like that, and now that I have a real engineering job I have realized this style is a lot more accurate to what the real world is like.

> Honestly I don't think I can fault students who cheated during the pandemic, it was just so abnormal that most people did whatever they had to to cope.

Cheating is never OK. Look what that lead to. A few bad apples ruined it for the other students. People have to learn, that their behavior in such a setting does not only affect themselves. In this case it very negatively impacted everyone studying with them.

Why are we constantly surprised when people focus on the only metrics they're judged by?

>People have to learn, that their behavior in such a setting does not only affect themselves

at some point it's like defensive driving. Sure, they are at fault and an idiot, but for the sake of your own safety you shouldn't "stick to your morals" or just "hope they get theirs". Look out for yourself, because for some people that's the only person that will do it.

With that lens, yes. I don't blame people who need to do what they do for one day in some artificial environment so they aren't impacted in their transition to real world work where they are otherwise moral.

> People have to learn, that their behavior in such a setting does not only affect themselves. In this case it very negatively impacted everyone studying with them.

Except, for many competitive positions, that's a zero-sum game.

Unfortunately you are right and it probably means, that for many competitive positions the process of evaluation is not a good one.

It's a grey area.

In india, the engineering curriculum includes lot of subjects that are not useful for the students but only there because some fat belly officer thought it would be "Interdisciplinary".

Can you imagine teaching biology as a subject for engineers? waste of money and not what we have come to learn, or interested in. But colleges in India do that.

Cheating in such subjects is fair.

It’s not a grey area, it devalues the degree for everyone and betrays you as a fundamentally dishonest and low value person.

Those subjects are not there for any meaningful purpose. Imagine having a CS degree program and teaching chemistry, physics, biology and a bunch of even lesser subjects is blatant and ignorant waste of students' money.

Tell me you’re an SDE1 without telling me you’re an SDE1

In engineering school, our professors wanted to treat things like the real world (as much as you could during test time).

Their solution was to let us use whatever we wanted except other students. Cellphones, textbooks, laptop, Internet.

You had 2 hours to answer 2 or 3 questions, it was definitely reasonable if you studied... But the questions were insane.

You know how homework would be assigned 1-50 odds? These were like the problems 74 and 76, 1/4 page long and difficult, multiple parts.

That seems as similar as engineering school got to the real would outside our sophomore and senior design projects.

> it was so frustrating not only the amount of cheating that occurred

Yeah, it's crazy that cheaters are the protagonist of the ban e-proctoring story.

What if the university only admitted students who don't cheat?

How do we know that policy works for real at a place like Caltech?

Academic culture is accommodating, and you feel that way too. But universities must know that cheating harms everyone, it's a losing proposition. The right response should have been to suspend grading.

Or just grade other work that doesn't require proctoring.

Personally, I removed exams entirely in the last courses I taught in Spring 2020, and it seemed to work out OK. I had no faith that the proctoring systems would catch everything, and I had no desire to inflict them on my students in any case.

> What if the university only admitted students who don't cheat?

It’s a common fallacy to attribute behavior to innate qualities of people, rather than situations.

For anyone interested, it's called the fundamental attribution error.

And yet there truly are people who are far more (or far less) honest than other people in the same situations.

I graduated in 2000, and cheating was rampant if you knew where to look.

One variety was legal - you’d have a doctors note to get a reasonable accommodation for ADD. In the 90s, it was pretty trivial to do that. Once you had that, you were taking tests in a facility that made cheating very easy (open book) if you were motivated.

Does anyone in academia know if tests are designed on a curve? You can give a test that everyone gets 100% on (assuming they show up) and you can give a test that everyone fails. Is part of creating a test thinking about how many students in your class are capable of scoring A’s, etc or is it just a side effect

I had a highschool calculus teacher who was very explicit about designing their tests on an A/B/C system.

The idea was that the A section was the basic plug-n-chug versions of the problem that everyone should be able to get if you were even half paying attention in class; the B section is stuff that requires a bit more reasoning and analysis, and may have a pitfall or two, but they'll be things that were at least mentioned; and the C section is effectively extra credit— a problem where you needed to have a deeper understanding in order to figure out something novel about it, or like you needed to re-derive something from first principles in a way that was slightly different from what was shown in class, that kind of thing.

And the sections weren't equally weighted; I think it was usually like 50/30/20, or even 60/30/10.

Some people hated this, but I really liked how explicit it was. There was no mystery meat— the whole evaluation scheme was very open about its design and intent.

From my experience... it varies widely. The university will have a standard, the department will, the professor will.

I had experience taking undergraduate courses at 4 institutions in the USA and of course many departments.

I had some professors that expected the grades to fall like grade school did with 90-100 % being a high percentage. Others gave 4-6 difficult multi part problems and graded on a transparent curve. Others gave difficult problems and were not transparent about the grading scheme.

Depends on country, university and professor. In Germany, the general rule is that grading on a curve isn't allowed or at least strongly discouraged. There was a short while when such a system was acceptable, called ECTS grades (not to be confused with ECTS credits), but that was done away with in a year or so. Currently, grades are based on the lecture contents and the expectation of the lecturer for the progress a student should make. Lecturers are relatively free to do whatever they please within those bounds. If a test ends in everybody failing, tough luck, maybe next time. If that unfortunate event does repeat too often, the dean might get involved, but can't do a lot more than talk to the lecturer.

If school is using grading curves that school has already failed and there is no saving it before grading system changes.

Not strictly curve grading, but when I was studying electronics one of the professors had a 9-step algorithm for processing raw scores into grades. That involved normalizing the lab scores between different reviewers (so that harshness of a particular reviewer would be compensated for). He also assigned `cos(percentage_from_other_test as radians)` instead of resits if you missed one of the tests for valid reason but had shown up on the other.

On one hand I really like the former approach and could understand the latter (organizing resits is likely a huge PITA). On the other that's the only professor I heard of that was attacked with an axe by one of his students.

> only professor I heard of that was attacked

How did that end, if it's not too weird to ask

> normalizing the lab scores between different reviewers

Sounds like a good idea to me :-)

Places that take academic honesty seriously tend to find ways to not have cheating happen in the first place. Caltech is notorious for it's take-home exams for instance.

But let's look at the reality of a student at a normal institution or trying to get into college today; it's a series of high stakes exams that decides your future. And a prisoners' dilemma since as soon as someone cheats you have to do it as well just to keep up. This used to be high risk high reward but now, thanks to the pandemic, it suddenly became low risk high reward. No wonders everyone is doing it.

Really, it's a great equalizer. Pre-Pandemic, it was a common strategy to get "diagnosed" with ADHD to get extra time during exams. But for that you needed well connected parents. Now everyone can cheat.

> Honestly I don't think I can fault students who cheated during the pandemic, it was just so abnormal that most people did whatever they had to to cope.

I beg to differ. Saying that dishonesty is acceptable in certain scenarios will only motivate them to further partake in bad acts. In extreme cases, the child will end up as a criminal.

"In extreme cases, the child will end up as a criminal."

these are university students pressured to have high grades for internships/first jobs, not some impressionable elementary schooler. Thinking way too obtusely for this to be a factor and compare this to some middle schooler getting into drugs.

Reply on news.ycombinator.com