Here are my remarks to the concerns below:

1. Time pressure
I know that this is a general problem with exams.
I try to make the midterm exam so that only a small number
of the students will make the
full score or close to it. This happened in the midterm where three
ladies got the top three scores.

My motivation for making the exam so that only a few can finish
is that I would like to let the stars ``shine''. The stars serve
as role models showing what can be achieved by the human mind.

The final exam is usually easier so that many students get
top scores.

2. Double penalty
I don't view this as an issue of double penalty; I view it more as
an issue of giving different weights to different unknowns.
I agree that sometimes it is arbitrary which unknowns get double or
triple weight. But often, I try to give higher weight when I think
an unknown is harder to find. Sometimes, my judgement might not
be accurate.

-- Karl Lieberherr
Dec. 11, 93

======================
Prof. Lieberherr

I have two concerns regarding the final exam.  

The first is time pressure.

I realize that there is a need for a time limit in an open book exam
or else some people would stay forever trying to look up the answers.
However, during the midterm I did not open my book once and I still
ran out of time.  The class dictionaries used are complex.  It
takes some time to assimilate them, say by drawing out a picture.  I 
think there should be enough time to do this.  In the midterm I did
well on Part I, but used up most of my time on it.  Hence, I did
poorly on Parts 2 and 3.  I hope that more time will be available
for the final exam.

My second concern regards double penalties.

In some cases, particularly the questions involving Demtraces, two
of the unknowns are obviously identical.  This is unavoidable without
giving the answer away.  However, I do not think it fair to take off
points in both places for a wrong answer.  Perhaps the grading method
could be adjusted so that this doesn't happen.  I know it is late 
for this time, but in future this problem could be solved by marking
both such paired unknowns with the same number. ex. UNKNOWN5 in both
places instead of, say, UNKNOWN5 and UNKNOWN10.