Sunday, September 28, 2008

Nonlinear Morality

I've recently become a guest of Harvard's Justice Online philosophy courses. A recent discussion posed the case of Dudley v Queen of England. The case describes a ship that sunk leaving the captain, a crew member and the cabin boy...maybe one the life boat drifting in open ocean. They had meager food supplies and no water. Soon they ran out and began to starve. The captain suggested they all draw straws and the short straw gets eaten by the rest. However, the cabin boy had already drank sea water and was very ill. In his weakened state, he became the victim of cannibalism to sustain the rest. The remaining crew was eventually picked up without having to kill and eat another member.

A cut and dry utilitarian perspective - the greatest good for the greatest number - suggests that their actions were completely morally acceptable. For objectivism, I've left out some details of the argument that would reinforce the utilitarian viewpoint. The question is: Is the utilitarian perspective correct?

The arguments for and against the case all have an underlying understanding of morality; one that is not so easy to define but rather somewhat universally understood. If it was, then philosophy would be objective and completely understood. The application of what essentially is the same definition of morality produces differences in option when different personal perspectives objectively weigh the means and ends of the decisions. We come to the first observation of nonlinearity. Morality is not deterministic. Presenting the same case to different people with always produce some variation in either the means or the ends.

Lets change the scenario. Only two people are in the boat and one eats the other to save himself. Is it morally objectionable? There is nothing known about either individual that might weigh on their worthiness to live. I think many/most people would find it objectionable. What if there were 6 people? One dies for five to live. The answer becomes less definitive. What if it is 21 people so 1 dies and 2o live? It becomes apparent that utilitarianism becomes more relevant once the prospering majority becomes proportionally larger. This is the second nonlinearity. Utilitarianism becomes disproportionatly relevant as the prospering majority becomes proportionatly larger.

I've already illuded to a conclusion I've made. Morality is not definable, yet is universal. It is the nondeterministic nature of moral application and interpretation that produces differing opinions on right and wrong. For example, in extreme situations we glimpse true humanistic nature in that self preservation becomes the driving factor and can be projected to apply to others. Self preservation is instinctual, primative, yet it allows us to justify extreme actions in extreme situations. With that justification we are exempt from the pain of guilt and remorse and utilitarianism still rules. However, in the case where only two people are involved and one is killed to save the other rather than 21, we can expect a level of remorse even though self preservation ruled; the second nonlinearity exists.

To understand morality I seems that we must consider human behavior on a much more primitive level. Even with that, I doubt that morality will be definable simply because of it's nonlinear characteristics

And now a deep thought...
If we consider guilt to be a primitive human characteristic, perhaps we must examine guilt before we can attempt to define morality.

Friday, September 26, 2008

Grading Classes

The other day I was thinking about the best ways to grade a class. I know that many of my fellow grad students plan to teach someday. How should we grade our classes?

As engineers, we benefit from the fact that we deal with a generally definitive subject. There are solutions to math problems, so we typically don't have to deal with wondering how to grade a specific problem. It is right or it is wrong. Partial credit can be given based on taking correct steps to get to the answer, but failing to get the right answer. In the end, this is all going to be much more straight forward than grading a literature class for instance.

I don't think they are as popular now, but curving a class is an option. I don't know if I have ever had an engineering class that was curved though. There seems to be some pros and cons. It does make it easier on the professor. Schools typically do not like professors to give all high grades, so the curve reduces this problem. It can help students if a classes or test is really hard, by at least benefit those who did well. Similarly it can hurt students if a class or test was easy, since it punishes the students who may have done well, and understood the material, but were slightly lower than the best students.

This seems to be the heart of the problem. What should a grade represent? In my mind it should be a qualifier of how well the student understands the material presented. Requirements for courses should be laid out prior to a class and students will be expected to meet them. If this is the case there should be no problem will all students getting an A, or all of them getting a F.

In the end it seems as if the best solution is to give appropriate tests. The test needs to be fair and represent the class requirements. If the test is sufficiently difficult it will limit the high scores, creating a sort of natural curve.

Homework and Projects play an important role as well. One system I like is having a few options for how your homework/test combination will lead to a grade. This gives options for students who may find tests nerve racking and difficult. Some people just have a hard time with tests. They can then focus on getting good homework grades to make up for the poor test grades. Additionally, students who demonstrate that they understand the material well by scoring well on tests would not be punished for scoring lower on homework. I always personally had this problem when I found a class easy since I did not feel the motivation to do the homework when I already understood it and my test scores demonstrated that fact.

The one problem I see with this setup is how to deal with giving a fairly graded classes on your first shot. It will be difficult to create a balanced test. I think the only option is to make it fairly hard. Afterwards you can at least curve the scores up. If you make the test too easy, I think there would be a riot if you curved the scores down.

So, what does everyone else think? How can we grade fair without making it a time consuming mess? What is the simple solution? What grading systems promote learning?