Banazîr the Jedi Hobbit (banazir) wrote,
Banazîr the Jedi Hobbit

  • Mood:
  • Music:

Behold My Wrath, or, Raaaaaaaaaask

I just spent two hours this afternoon going over AI homeworks with two of the graduate students, one of whom is decent and the other of whom is quite good.

What the crap is going on with our graduate program?

Has the whole world become a haven of intellectual lightweightedness? The people who came to see me from 14:40 through 16:40 are intelligent, talented, knowledgeable, and generally hardworking students. Five or ten years ago I would have given them twice the work they are doing in my class and expected the same level of quality. Now?

Let us sit upon the ground, and tell sad stories of the waste of minds.

"I already have an exam that day."

The opening volley came last week, when a student pointed out to me that he had two final exams on the day of mine (Monday, 13 December 2004). He waited expectantly while I asked what he expected me to do about it. "Nothing," I replied, "unless you have a third, which I can excuse you from. It is three exams on the same day that you can't have." Blank look. "Do you?" Unrecognizing stare, slow shake of the head. I bit back my true response and remarked that undergraduates with full 15-20 hour course loads often had two exams on the same day, including two upper-division (700-level) courses. "Oh, in [my home country], we always had two days in between."

Well, guess what? Tough cookies, buddy! Cry me an inland freshwater river!

Now, those of you who know me know that I am not a Heartless BastardTM, though I have been working to reduce my reputation for having a soft heart and shore up the high standards I allegedly have. (They aren't high; see below.) But people? Since when is it appropriate to ask the instructor for an exam exemption - possibly necessitating an alternate make-up version of the exam - because you have all of two exams on the same day, boo hoo? Is it all just about what you can wheedle out of the instructor now?

"I had no idea graduate teaching assistants could be assigned more than one course."

Dude. When I was a graduate student (for that is how good rants begin, and you know it), I worked for three semesters as a teaching assistant:

  • Spring, 1992 - at Hopkins, as a grader for 600.140 (Programming Techniques, a "CS1.5" course using Abelson and Sussman that would go between CIS 200, Introduction to Computer Science, and CIS 300, Data Structures and Algorithms, here at K-State)

  • Fall, 1994 - at Illinois, as a half-time TA (0.25-time by K-State measurements) of 1.5 TAs (0.75 TA units at K-State) for each of CS348, Introduction to Artificial Intelligence and CS325, Principles of Programming Languages - analogues of CIS 730 and CIS 505/705 here. There were 125 students (dropped to 115) in CS348 and 125 (dropped to 110) in CS325.

  • Fall, 1996 - at Illinois, as a full TA (0.5-time) for CS125, Introduction to Computer Science. There were four TAs, 375 students at the start of the course and 250 at the end, and 14 weekly hour-long lab sections of which I taught 3 plus and substitute labs.

In the first job, I worked for 15 hours a week and put down 10. In the second, I worked for 15-17 in each course, and in the third, I worked for 25-35 hours.
Complain? Heck no!

No idea grads could have their duties split across multiple courses? Let me get this straight. You had no idea that M.S. and indeed Ph.D. students could be assigned by "one or two tenths" (units of 4 hours per week) to TAships? Welcome to the real world! Back in my day, we worked. We held office hours, wrote homeworks, graded all the instructor's assignments, posed and graded exam problems, taught labs, proctored exams, and gave nighttime emergency and remedial help sessions (out of the goodness of our hearts, from 19:00 to 22:00 some nights). And we frelling liked it, you pansies.

"You said this would take one to two hours. It took more than four..."

Let me make this clear: I hate this. I despise, deplore, abhor and otherwise heap general distaste on it. There is almost nothing I resent quite as much as having a time estimate that I confirmed by actually working the problem challenged. Did I actually produce a model solution? Yes. Would you, who had to install Eclipse and check sources out and download things from SourceForge, need more time than it took me? Indubitably. Count on it; deal with it. Am I faster than you, a student who is still learning how to solve the problem? With high probability. Know that I took that into account. Did I know more than you did coming into this course, to be able to solve the problems I wrote or assigned out of the textbook? Perhaps. (Possibly, but not necessarily.) Am I still capable of underestimating your learning curve? Almost surely. Will I take it ill if you hold me accountable, not for underestimating my solution time or not planning, but extrapolating to you too optimistically? You bet your sweet bippy.

Seriously, folks: this is one of my very strong pet peeves. IMO, instructors are responsible for:

  • Setting a reasonable workload. Arguable in my case, but see below.

  • Not assuming that students have no other courses. Listen up, people: when my father was an instructor the professors used to declare as a matter of policy (self-deluded policy, IMO) that "you have no course but mine". Now obviously this kind of nonsense is not going to wash "in this day and age", but I mention it as a data point: there was a day when 20 hours of work outside lecture per week was not unusual for a 1-unit (4-semester hour) course, and people still survived, graduated with Ph.D.s, got faculty jobs and went on to craft the pansy-friendly world you young folks live in today.

  • Being prepared for lecture. (OK, this one I have no excuse for. I wrote out dozens of PowerPoint lectures four years ago and have been very slow about revising and updating them.)

  • Delivering clear and well-paced lectures. (I think my lectures are clearer than they used to be, but apparently they are still too fast.)

  • Grading assignments conscientiously and giving feedback in a timely fashion. (This is a whole other rant, coming soon)

  • Being helpful: accessible to students, reasonably willing to meet, etc. (This has got to go in the same rant as above.)

Nowhere in the above does it say that we are responsible in the slightest for anticipating the idiosyncratic difficulties caused by lack of background, motivation, or simply a can-do attitude on the part of each and every student.

"I didn't know you expected us to read the next page."

Dude. I am here to teach you some new concepts, to help you see how to relate and apply your previous background in computer science, mathematics, statistics, cognitive science, neurobiology, etc. to new problems. I am not going to paste exact paths, each and every link you should be reading, the start and end pages of reference literature. You know how to use Google,, and CiteSeer/ResearchIndex; you should also know how to use Current Contents, Books in Print, maybe PubMed, INSPEC, COMPENDEX, etc. If you don't, there are highly capable people such as sui_degeneris and myng_rabbyt who are able to assist and direct you. They are called research librarians. You know how they have a lot of training? They had to acquire (yes, acquire, as in go out and get, not sit and be fed) a considerable amount of knowledge. Knowledge that you are also, as graduate students, expected to acquire, through a process called ed-u-ma-cation.

When I was an undergraduate taking Simon Kasif's Intro to AI course (600.335, later cross-listed as 600.435) would assign maybe two of every five problems out of the textbook (Nils Nilsson's Principles of Artificial Intelligence). The rest of us would go to the Milton S. Eisenhower Library (MSEL) in the early mornings (before 09:00), late afternoons, even nights and weekends, to garner a few fleeting ours not on workstations, but on external references not in the reserved book list for the course. Interlibrary loan (off-campus borrowing) was our friend at JHU and UIUC. Nowadays? How many of you, who were not told by me, even know where the book purchase recommendation form is?

Here, have a free clue, locals.
(You folks at other universities have these buildings, too!)

Pardon my French, but it's quite enough. We, the faculty, are here to facilitate, not to turn pages. Teachers: providers and guides to resources. shu1 tong2 (scholars' assistants): not so much.

"I didn't get back any of graded homeworks 2 through 5."

Uh-huh. OK, see, now that's where you stop right there and check yourself. My grading is just a tad bit slow, which is why I have a policy: the next homework will not be counted late until the previous one has been graded and handed back. Now, what does that policy tell you? That I intend to get assignments graded before the next ones are due, to avoid losing an assignment to schedule slippage, probably. More important, though: it should tell you that I, personally, would not dawdle in getting scores and score sheets for graded assignments back to you. Let me be clear: If you are telling me you are missing more than one assignment score, you have probably done something wrong. To wit: unless previously announced, there should not be more than one extant assignment that students are working on at a time, nor one extant score being handed back at a time. In other words: tell me early (as soon as you know) if you are missing something. Most especially do not pronounce to me in a snide tone of voice, "see, now it is harder for you and me to remember [the details of this problem set]". Understand: that ship will not sail from my dock.

"I came to your office and you weren't there."

Again with the self-checkingness. Here, please read this 2002 rant; it will tell you everything you ever wanted to know about getting in touch with me. I am up to 600 e-mails a day now, of which 90-95% are easily spam, but the recommendations in that article are still valid. (In fact, the only reason it is no longer linked is that I wanted to reformat the HTML that masaga produced as I did with this page.)

What does this say? It says you are responsible for forwarding and checking your mail. (This problem is endemic in our department, so my remark is not meant to single out any individual.) You are furthermore enjoined to read my schedule, which is linked from my home page, posted as this snipped URL on my office door, and otherwise archived by our department student secretaries.

"Your standards are impossibly high."

Utter nonsense. I checked myself, so let's have self-checks all around. First of all, it is true that grades do not matter much at the advanced graduate level: I rather like the scheme kakarigeiko's university uses, wherein all Ph.D.-level coursework is graded pass/fail. Perhaps even better would be the "real grade" (the cold, hard truth) and "ironic grade" (inflated grade for transcripts, grad school applications, and mothers' refrigerators) that Harvard professors bandied about as a bright idea that ended up being quashed. I know I sat through Gaisi Takeuti's linear logic course and didn't learn nearly enough, even in later self-study, yet he gave me an 'A' anyway. So did Nachum Dershowitz, who gave "self-assessed" grades ("tell me what you think you deserve"). On the other hand, I audited David E. Goldberg's genetic algorithms course and P.R. Kumar's information theory course. I did a lot of the work, yet those courses appear only as audits on my transcript. In the end, I kept a 4.0 GPA through all five years of my graduate program. So what? Grades only mean so much. It is the effort and the knowledge reward that they reflect that means something, not some marks on paper or bits in a system.

So, here is why I don't like the notion that my standards are unrealistic. I frequently mention Structure and Interpretation of Computer Programs as a fantastic CS1 book. I say it often enough that I'm sure it gets on some people's nerves. At Illinois I was told, "this isn't MIT". At K-State I am told, "this isn't Illinois". Let me sum up: Just saying that is promoting a myth. Is Hopkins on par with MIT? Is it on par with Illinois, for that matter? And yet Amy Zwarico used SICP to devastating effectiveness to rout the monster that is CS ignorance. I love SICP with a passion, I will give it as a present and beg students in the first two years of a CS program to read it, while the neural plasticity is still there, and I will unabashedly stand by my assertion that I care not whether I am at University X. A good book is a good book, whereever it be read; a proper standard, a proper one, whereever it be adhered to.

Also, let's get this straight: MIT became "MIT" (i.e., established a worldwide reputation for excellence) because they dared to set and keep the high standards, to not let themselves become anything less.

"If you held your teaching assistant to the same standard you hold us to, he wouldn't even get a 'D'."

It's an invalid premise: I cannot really hold sample code given for functional convenience to my grading standard to begin with, much less to my standard for extra credit problems as was the context.

I said as much, and the answer was "so, did you give an 'A' or a 'B' grade to the MapGen solution"? I explained that the standards applied to Ben Perry's solution in 2001 and to the work of hpguo (2000), masaga (2001), scottharmon (2002), and Charlie Thornton (2003) were the all the same as in 2004, and got a sarcastic "interesting" for my pains.

Well, I didn't take kindly to that. I had asked masaga to cobble together a "crap of a crap" sample on short notice, and people knew this. There aren't enough Caveat Lector disclaimers in the world for that code. What was it masaga said? Ben would weep if he saw it... well, that's as may be true, but an "as is" disclaimer (or at least a "this is not `model code'" one) might have headed this attitude problem off at the pass.

Moral: Don't nitpick - especially when you aren't sure what you're talking about.

"I would like to see model solutions so I can see where I went wrong."

Let's get one more thing straight for now: there is, in general, no such thing as "losing points" on my homeworks and exams. I do deduct points for common errors, but generally I award points. My father taught me this one: he would often quote the instructors at his undergraduate university (Taida or National Taiwan University) as saying that they always thought of points as being given rather than deducted, because if the scale were based upon point deductions, there would be a large incidence of negative scores.

And while we are on the subject: postgraduate students should really break themselves of the "grades as feedback signal". Any graduate student should be able to maintain a high GPA and score well by testing well. See my remarks on grades above. Furthermore, you who are in graduate school should have started to learn to become more active learners. "I only know what I am doing wrong if you tell me" is, in general, not the right attitude. I know because I had it as a first-year graduate student and was broken of it very forcefully. Finally, grads should know better than to harbor a "your mistakes are your mistakes; my mistakes are your mistakes" stance. They can deny it if they will, but I have so often seen the "how was I to know?" attitude extrapolated to "lead me, feed me". Guidance and facilitation is one thing; but graduate scholars are, at the heart of it, knowledge cultivators (or foragers), not mere knowledge consumers.

  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.