Hacker News Reader

Comment by robobenjie | Hacker News Reader

My school (Stanford) the teachers were required to leave the classroom during tests and we were supposed to use the honor system. I never cheated and I don't know anyone who did. Maybe his sarcastic comment at the end has more truth in it than he meant. People tend to live up to expectations.

If you prevent cheating by having a person who's job it is to catch you then you feel like if you don't get caught you have won. On the other hand if you are expected not to cheat because it is not what a honorable person does, then if you cheat it is because you have failed.

View on HN

I was a TA at your school (Stanford). I like that policy, but I also witnessed cheating firsthand. Specifically, I graded a take-home midterm exam where 3 students handed in the same bizarrely wrong answer. The students received penalties on that test, but no other disciplinary action was taken. This was in the mechanical engineering department in a graduate level mechatronics course (that I suspect you've taken).

If I were making the rules, those students would be out of the university the same day. There is no shortage of smart people who would love to attend Stanford, especially the engineering school. I have no idea why enforcement is so disgracefully lax.

In the electrical engineering department, I regularly had classmates ask me if we could "compare answers" on homework problem sets. Sometimes (rarely), this was for a genuine goal of education-- if your answer is wrong, you want to know that so you can figure out the right answer. More often, it was clueless people trying to collect answers from multiple other people. Drove me up the wall.

In my experience (another school, another country), it was usually allowed to take a take-home exam together with others. The understanding was that the exams are sufficiently hard, that students smart enough to pass do not allow others to freeload and that working together with other smart students teaches important skills in addition to a better understanding of the subject.

The same goes for homework problem sets: me and two others would usually first solve them ourselves to the best of our ability and then compare answers, which was always an educational experience; the more so when no one had successfully solved the problem on their own or when it took someone a while to convince the other two their (identical) solutions were wrong.

> The understanding was that the exams are sufficiently hard, that students smart enough to pass do not allow others to freeload...

That sounds like a flawed assumption to me. There are lots of social and economic incentives that might lead someone smart to do others' work.

I tend to agree that working with others on problem sets can be quite beneficial. Many classes I took had the same policy on those. But at some point you have to prove you can do it yourself.

I find that comparing answers make a lot of sense when you all know each other and your level of skill. I often check answers with my roommate on math problem sets, and I know we're about at the same level, so neither of us freeloads.

It would be interesting to have a web application that both people submitted the answers to and it came back with just "same" or "different". Might be one was to genuinely compare quickly. Then based on the number of sames you could discuss the differents (or decline to).

My first year at Caltech, where almost all exams are take-home exams, I was in trouble. I'd spent way too much time at the computing center hacking, and at the end of second quarter I'd had to petition to be allowed to stay. I was allowed to stay on the condition that I pass everything third quarter.

I flunked the final exam for physics. I knew I was flunking it as I took it. I also knew EXACTLY where in the book the one bit of information was that I needed to pass that exam was. All I had to do was reach over a couple of feet, in my private, closed room, and spend 30 seconds in the book, and all would be well. But Caltech was on the honor system, and the instructions on the test said closed book, so I flunked and was kicked out.

(I then spent a year at a community college, acing everything, to show Caltech that I could actually do college work if I tried. They let me back in, and enough of the credits from the community college transferred so that I was able to come back as a sophomore. I still ended up spending way too much time hacking, but ended up with a B average at graduation).

I hope you are responsible for vital systems or research. You are demonstrably a person of character, head and shoulders above the rabble that is usually put in charge.

Hmm. I'm unethical then. Here's my reason why I would have read the book and passed (and saved money for that class).

Any money borrowed for purposes of education in the United States is unattachable. That debt will stick with you, NO MATTER WHAT. We have Reagan and 98' Republican Congress to thank/blame for this. The only other debts that have this status are ones granted by a judgement for crimes (for example, murder).

As I am reminded by that art student who had 160K in debt that was posted a few days ago, it is fiduciary irresponsible of you to have that kind of situation possibly happen.

> We have Reagan and 98' Republican Congress to thank/blame for this.

What's your response to the problem of folks declaring bankruptcy to get out of student loans right before they start earning real money, such as MDs?

> As I am reminded by that art student who had 160K in debt that was posted a few days ago, it is fiduciary irresponsible of you to have that kind of situation possibly happen.

Which "you" is that about? Are you saying that art students shouldn't be given such loans or they shouldn't accept them?

I think the best way to prevent cheating, at least for in class timed tests, is to make them non-trivial.

Multiple choice, simple fact recall -> easy to cheat

Real idea synthesis, non-trivial 'show your work' calculation required -> hard to cheat

It comes down to amount of data that needs to be moved surreptitiously to mimic a correct answer. When you frame it that way it almost looks like a problem in the security domain (secret keys, side-channel data leak, time bound attack)

Unfortunately it's also the case that: simple fact recall -> less need to cheat. Non-trivial calculation -> more need to cheat.

There are other reasons to not use simple fact recall type tests, of course. Like actually testing transfer.

But, in support of what you said: I went to Uppsala University in Sweden, and all exams were 6h long and happened in a giant special-purposed exam hall with 6 feet spaces between each desk. Cheating by communicating was a lost cause and even if you could, the amount of info you'd need to transfer to make a dent in a 6h-long test would be way too large. I don't recall anyone trying to cheat, but if it was to happen, it would have been done by adding notes to the bible-sized book of tables and formulae that we almost always were allowed to use.

> Real idea synthesis, non-trivial 'show your work' calculation required -> hard to cheat

And also harder to evaluate and grade; the professor or TA must be willing to invest an hour or several in reading through the longer answers to see where the students went right or wrong. That would be no problem in a hypothetical ideal educational framework, but realistically TAs (even at Stanford) aren't always that dedicated.

True, but maybe that means you just give fewer tests that matter more.

> "If you prevent cheating by having a person who's job it is to catch you then you feel like if you don't get caught you have won."

Given the time of year, when I read about cheating and the honor system, my first thought is about taxes, specifically, paying state sales tax on items bought online and shipped from out of state.

Most states don't have an easy way of catching sales tax cheaters, so it usually comes down to the honor system: When you file your state income tax return, you are supposed to add up all your "mail order" receipts for the year and pay sales tax on them.

How many HNers actually do this, I wonder? This seems like an instance where there is an honor system, but everyone cheats anyway.

The internet sales tax honor system is more of an oner system. You have to go out of your way to do all the bookkeeping yourself to reward unwarranted rent-seeking.

I went through the online H&R free tax thing because I made less than $50,000 in 2010, and the entire way through I am looking for a place to add up my receipts and pay my state tax on them, nowhere to be found. I have no way of calculating what else needs to be withheld, I don't have the money to go to a tax person to get it done there.

Tax law has become so complex that a lay person can't get the job done without either paying someone to do it, or having to use software that may not include all of the options required...

I also went to Stanford and when I took a beginning level CS course a student in my year was caught cheating. He copied a problem set from someone else and was later suspended for a quarter. Stanford has a judicial review board composed of students that make the final decision in cases like these and he was still punished. Everyone was warned by the professor at the beginning of the class that the CS department catches the most students cheating every year b/c every problem set is electronically reviewed for ways it might have been copied.

I went to a prepschool with a strong honor code culture. Cheating was nonexistent at the higher level classes I was in but it was pretty rampant among those in less intensive class distributions.

I think the deciding factor in successfully inculcating the value of fairness in academia is ensuring the students actually value academia. If the work's viewed as strictly an impediment by a student there's almost no way to change that.

I think it really depends on the students and whether or not you are finding these coasters (taken from other comments) in the classes or not. Cheating is probably the least likely in the harder degrees near the end of the program and most likely in the easier degrees near the beginning of the program.

In my engineering program I have had plenty of teachers walk out of the room for large amounts of time or do something like grade papers without looking around the room and haven't once seen an issue of cheating.

However in my freshman chemistry course which was required for a large amount of students from many disciplines they had 8 T.A.s and electronically moveable cameras (the class was recorded so the cameras were just being re-purposed) monitoring a class of about 200 and I do believe some cheaters were caught in my semester.

I was in the grad program at a top 5 stats program. Cheating was rampant. I discovered this in a 2nd semester grad level math stats course. There were 4 problems on the test, and one of them could only be solved either by being much smarter than me or by getting kind of lucky and very quickly divining the right, tricky way to factor the product of some characteristic functions. I thought it was very non obvious. When the class discussed the test there was a bimodal distn of scores on that question; 3 students, including me, got 1/2 credit and everyone else got 0 or full credit. I asked a classmate how in the hell he figured out the problem in the very tight time limit for the test. He basically told me that last year's test -- not distributed by the professor -- had almost an identical problem on it. Once you'd seen the right way to factor the characteristic function the problem was trivial. Pretty disappointing.

Forgive my statistical ignorance, but isn't this form of cheating easily remedied by substituting a different class of characteristic functions?

If a different class of functions doesn't exist, it seems like your classmates just had better study material.

(I'm thinking about a calculus test with a previously unseen e^x. Wouldn't knowing the answer from looking at a previous year's test be more akin to 'learning' than 'cheating'?)

Lots of organizations on campus (e.g. greeks) collect tests from their members and file them. This "helps" their members do better on tests. The situation you describe is basically one of the outcomes of organization resources vs personal resources in higher education.

Sometimes, it can work against the student. Particularly if the student doesn't attend class (low-level blow off class) and the teachers / method changes. I TAed an intro to programming course for non-majors. My fellow TA and myself created new tests for the class (it went from GWBasic to QBasic). A couple of students from one of the fraternities attended the 1st day of class and then the first test (missing the homework programs). I handed out the first test and got the following reaction: "This isn't the right test!". Oh well.

Reply on news.ycombinator.com