Most of the marking of assignments in CS courses is done by graduate TAs (teaching assistants). In the first two years, ISG full-time undergraduate tutors supervise them; in senior years, there may be a head TA, or the instructor may simply distribute assignments with marking schemes. It is rare for instructors to mark assignments, except in very small fourth-year courses. I'm going to talk about TAs in general and not the specific ones associated with any course I've taught, because I actually have had very little exposure to the latter.
The TAs work for either five (one TA unit) or ten hours a week. They are promised a certain number of TA units when they are accepted to graduate school. Their stipend consists of TA units and RA units (research assistant funds, paid for out of a professor's research grant). Together these give graduate students enough money to live modestly and pay their tuition.
Some TAs do a good job. Some TAs neglect their work, marking hastily and with little understanding of the course. Though they are supposed to have completed, in their undergraduate degrees, work equivalent to our required CS core courses for the B.CS degree (and some have to take extra undergraduate courses to address this), the assessment is done based on their transcripts and their written descriptions of their educational history. They may have covered a subject but in relatively little depth, and they may not have used the specific language, notation, or textbook used by the course they are marking for. Other TAs are too conscientious: they mark slowly and carefully, using up their hours before the work is done, or getting little done in a group marking situation.
Because there is a support component to TA salaries, they are paid above the market rate, probably about twice as much. That is: half the money they earn is actually for the work, and half is money coming to them for support. Consequently, there is a certain disconnection between the work and the reward. If a TA does a mediocre job, not much is going to happen to them; if they do a poor job, they will probably be assigned to a different course in the future. If they don't do their job at all (not showing up to tutorials, or chronic lateness in returning marked assignments), they can lose their TA unit, but this is only done in extreme cases.
It would seem to make sense, from the point of view of improving TA quality, to split the support and work components of a TA unit apart. People are reluctant to do this. It could make grad students less willing to do the work; they will make do with the support alone, or convince their supervisor to increase their RA funding. Is it better to have a few reluctant workers, or have them drop completely out of the system, leaving professors to reduce the amount of feedback given to students and increase the weight of examinations?
In CS 341 (Algorithms), I typically find that about half my TAs are good; they are intelligent, mark effectively, and genuinely wish to help the students. The other half are less effective, in various ways: they may be weak on the material, they may be arrogant or unhelpful towards students who approach them, they may be slow, they may try to get away with less work. (It is very rare to have all of these occur at once.) Lower-year courses are less demanding; I have heard that many graduate students try to get assigned to first-year courses, where the marking is relatively easy. (In fact, these courses are used as training grounds for incoming TAs, about which I am also ambivalent.)
TAs often feel that they are caught in the middle. On the one side they have professors who may routinely assign marking duties incompatible with the number of hours they are supposed to work (this is reportedly a major problem with the OS course, for which very few TAs volunteer) or design assignments in haste and leave them to fix up problems and work out solutions. On the other side they have students who take them for granted, plagiarize work, write illegibly, and demand that marked assignments be handed back in an unreasonable amount of time.
As with so many other aspects of university life, this is a situation where the status quo is open to criticism, but the first (and perhaps biggest) problem is a lack of understanding of details of the situation, and a lack of communication among the various people involved. --PR
(Adapted from a blog posting made November 27, 2003.)