# Machinations

Homework and the Internet
January 29, 2010, 8:56 pm
Filed under: Uncategorized | Tags: ,

This is what I have been reduced to:

“Let P (x), Q(x), R(x), S (x, y) be the predicates, “x is a true dungeon master”, “x has sex appeal”, “x is a wood-elf ”, “x is a friend to y”. Translate the following statements into predicate logic.

• A true dungeon master is a friend to all wood-elves
• Only true dungeon masters have sex appeal
• Bob is not a friend to some wood-elf

Prove from the statements that Bob does not have sex appeal.”

I’m not sure when it happened, but for at least several years it has been the case that solutions for the problems in any good textbook can be found somewhere on the Internet.  When I first started teaching, my homework sets were usually a combination of easier problems from the text book and harder problems that I made up.  These days, even for the easy problems, I can’t really assign anything from a text book.

Definitely the ubiquity of solutions to written hw problems has its downsides.  It has made my job a little bit harder, but this hasn’t been a big problem really.  The major downside is that there some very good problems that students are missing out on.  For example, I recently covered the proof that square root of 2 is irrational in my mathematical foundations of computer science class.  A great hw problem to ask students after they see this proof is to prove that square root of 3 is irrational.  However, I really don’t feel it is fair to assign this hw problem in a large class where certainly at least one student will look up the solution on the Internet.

I’m curious how other lecturers are dealing with this issue, and how students feel about it.  Should we be trying harder to hide the solutions to good hw problems?  Should we plant “fake” solutions on the web?  What happens when you google “dungeon master” and “sex appeal”? (P.S.  Hello to my CS261 students!)

The thing is, being able to seek out previously-existing solutions is actually a pretty big and valuable skill. But you are right that it is a bit frustrating in fields where the traditional homework problems are exactly that – traditional. I think the solution is to ask bigger and hard and more open-ended problems. It’s a lot of work to come up with HW problems, so we should only give out a few each week and make them individually require a lot of work.

In algorithms, for example, I made students find an algorithm for a problem, prove it correct, analyze its timing, implement it, and feed it test data of varying sizes to verify that their analysis was correct. This was a lot of work, but it also was not really “cheatable” the way that, say, “Prove sqrt(3) is irrational” is. I have less and less-good advice for a discrete math course, largely because I haven’t taught one.

If we don’t want our assignments to be Google-able then they need to (a) not have the same proper nouns as the real problems and/or (b) be off the beaten path. The second is, of course, much harder than the first, particularly in math, where there are more paths that have been beaten for longer.

Comment by Peter Boothe

Peter, I agree that Internet research is a pretty valuable skill. I do try to assign problems that encourage use of things like mathematica or maybe reading parts of research papers. Sometimes it is possible to take a well known problem and make it Ungoogle-able by just changing bits and pieces of it. However, there are also some really good problems where this is just not possible (i.e. prove sqrt(3) is irrational). I guess these will all just become “at-home exercises” in the future.

Comment by Jared

I’ve been influenced by Jeff Erickson in my thoughts. In my grad algorithms class last quarter, I asked the students to not use the web, made the assignments challenging, allowed them to work in groups and made the assignments the bulk of their grade. This was what my classes at Brown were like. Even the finals were take home. I regretted my choice.

Now, assignments are worth little. I don’t make a point of saying “no web searching”, tell them to cite any references they use and will, of course, come down on those that plagiarize. If they don’t put the work in, they won’t do well on the in-class tests and final exam.

Comment by Glencora

Glencora, this is interesting – I’m doing sort of a mix of these two approaches. I make the hw assignments worth very little but I tell them that they can use any reference (including the web) so long as they cite it. I have recently started to give some take-home midterms and finals that are also “open Internet”. I spend much more time designing the questions on these so that the Internet will not really help them. I definitely like take-home tests because I remember learning a lot from then when I was a student.

Of course the danger for the take-home approach is that they will cheat on the exams by working together. However, I’m a bit less worried about this because it seems that for pairs of students who copy off each other usually both students are not that good (at least the ones we catch). I also usually find a much better normal distribution on the test scores than on the hw scores. I may go back to the in-class tests if I find that students are cheating in some way on the take-home’s that I can’t yet detect but I’d prefer not to.

Comment by Jared

[…] week I read an interesting article about homework and the Internet (shared by Lucas Wiman). The gist is that the Internet has trivialized homework  because students […]

Prof. Saia,

I’m not sure about the google search, and I don’t think you should post ‘fake answers’ on the web. I personally am somewhat disappointed to hear that we’ll miss out on some classic problems (like the irrationality of sqrt(3)). I’m not quite the type to solve such a problem purely out of my own interest, but having it as a homework problem would be a nice transition from the lecture, with enough inherent challenge to make it worthwhile. With regard to students cheating on the internet, I’m a bit cynical. I figure that students are responsible for their actions, and if they choose to cheat, it will hinder themselves in the long run (in later classes like algorithms) as well as in their future profession. They can learn the easy way or the hard way, but I do hope that motivated students aren’t spared some meaningful assignments/problems due to bad intent of the rare few whose decisions will only hurt themselves anyways.

Cheers,
J

Comment by J

Jody,

For the most part I agree with what you say. However, as an instructor, I want to protect those students who are following the rules against those who are not. If there is an honest student who is on the border of passing the class, I don’t want cheating students to artificially move that border.

Comment by Jared