dubious.biz(ness)

3/4 of the way through exams. Computer is dead. Have been ruined by the Internet. Specifically: I’m looking at comics. Web comics. The sort of comics which have absolutely no requirement to be funny and contain only jokes which in-reference other stupid internet cultural icons. Just imagine if Charlie Brown wore a different smurf t-shirt every day and “accidentally” mistook people for other the characters in adjacent comics on the funny pages. Now give him glasses and make every other frame have a paragraph’s worth of text in it. That’s internet comics. That’s rock bottom.

Anyway, I’m almost done; my brain is wasted. The frustrating thing about these exams, the thing that I have been frustrated with since I realized it last year, is that they are designed around such ridiculous premises that despite the examiner’s best efforts they are so broken that everyone games them completely. Why?

1. Relative ranking: Exams are marked on an absolute marking scheme and then people are “classed”, which could be considered a type of curve in which one defines that level will contain a certain percentage of the entire class and then fill those numbers. The boundaries are not absolute (though they are determined by a lengthy procedure which is well documented).
2. Absolute question scores: Each question is marked out of 20 points, usually with multiple parts, each being assigned a number of points individually.
3. Multiple options for questions: Papers this year have 9 or 12 questions, and you must submit answers for 5. There are some other limits (so many from each section) as well but the end result is that every student is not answering the same questions.
4. Few questions: There are only 5 questions answered by each student in each paper and there are only 4 papers, so each student submits 20 questions. Admittedly each question is multi-part, and they are not easy. Quite the opposite, they are very difficult. But there are only a few so relatively speaking each one is worth 1/20th of your grade.
5. Extremely difficult: One of the most notable things about the questions is that they are all very very hard. Many of them not because they require in depth thinking but they require complete mastery, or at least memory, of the course you have taken. On average, a question will cover about 15-30 minutes with of lecture (the same time as to answer it). However in the year I have attended about 20 lecture hours per week for 24 weeks. So about 1/24th of the material is actually being tested.

I can explain a number of these in detail, but I won’t. I actually do, to some extent, understand the system. It is motivated by the Cambridge culture which has immeasurable inertia, the fact that most students are brilliant, most lecturers would rather not grade papers, and finally ironic attempts to patch up the system without ditching it. I complain, but other degrees in Cambridge are essentially the same and can be much worse.

The real problem is this:

Considering 1, I know that no matter how I do I only need to do better than my peers to do well. It is a symmetric game, where my gain is roughly someone’s loss, but only if I lie near the borders of a class. Otherwise we’re likely to do about the same since everyone will break their necks to win.

Considering 2 and 3, it is in my best interest to choose the questions which are likely to gain the most points. There are a number of ways of determining this beforehand, but suffice to say that just a challenging question with a leinent marker who considers “effort” to be sufficient is better than a challenging question with a strict marker who only takes the correct answer, while an easy question under the same measure is exactly the same. While I can tell which questions are easy and which are hard (or rather which I know and which I don’t), I cannot tell the type of marker. And since not everyone chooses the same questions, I could end up with an evil marker who marks my very good but wrong attempt at a question low while others end up getting full or partial marks on a different question with a different marker with better disposition.

Considering 4 and 5, I go in to each exam with the knowledge that a) any question could occur so I must know everything and b) that most of the things I know will not be examined. Another way of saying this would be I waste a lot of effort commiting things to memory that I will not use again. However because each question is worth so much, because there are so few questions, I cannot afford NOT to know everything.

Why is it wasted effort? Simply because I am not expected, in my real life, to without reference the things I am expected to know in the exams. The exams do not reflect real life. What is more, not only do they not reflect real life, they do not award the most academically sound nor the most intelligent, they award the most dilligent with the best memory. For a high level institution to put such low emphasis on academics, after a full year of “getting you to think beyond your abilities” which I would say Cambridge does absolutely do, to go in to an exam and be requested to regurgitate formulas on from page 65 of the handout which were discussed for 5 minutes, which are worth 1/50th of my entire year’s mark is insulting. To make it even worse, because I know that the range of classing between the best and the worst is about 10%, that’s actually 1/5 of my grade from a third to a first. So if I fail to recite more than 5 such formulas on command, I’m a dunce, while if I can do all 5, I’m a genius.

This reminds me of the sort of stupid “IQ” test you get in email which has a scoring rubrik comprised of “15-20 ALBERT EINSTEIN!!1! 10-15 Highly Intelligent 5-10 Average 1-5 Go have a nap you dipshit”. If the test were meant to show how academic I was, I’d be happy to fail. In fact, one of the questions I attempted asked something that wasn’t anywhere in the material but could be proved with some ingenuity from the tools we had learned to use. I answered it wrong (or at least not well). Do I feel bad? No, the question bested me, I couldn’t prove it and I did what I could. However, what does frustrate me is that while I chose that question, others chose a question which asked them to recite a (rather long and difficult) formula they had memorised. If they could do so they would get the full 20 points, no questions asked.

That is frustrating. It’s the combination of random chance and intellectual insult that really pisses me off about Cambridge, because I know for a fact that I’ve never worked so hard for something so stupid in my life as this degree, and I sit in examination, with full knowledge that there are literally 100 ways I could obtain the backpropagation learning formula for neural networks, not the least of which involves looking it up on a computer. And I suppose the counter argument is that if I really understood the backpropagation algorithm, I would be able to recite it. I find that ridiculous as I fully do understand it, it minimises error by a finite amount by recalculating weights…blah blah blah. I don’t expect that unless I was in the fairly boring field of neural network implementation that I would ever need such a formula. But if I had, I might have a first. Likewise if I had known the two theorems of welfare economics… etc etc.

At least with rowing, I know when I’ve lost, it’s because the other crew was measurably better. But alas that’s another story.

My damned computer has decided to completely give up the ghost quite abruptly. The problem is that the backlight on the LCD is flickering on and off and then it just blinks out. It appears to be a problem with the backlight itself (which I presume is both florescent and has a ballast), as under very strong direct light I can still see the screen image after the backlight has gone off. Sometimes it causes the machine to crash, other times I can just try to suspend and “reanimate” it several times until the backlight comes on and stablizes, much like a nearly dead florescent tube. This trick worked for about a week but now it’s given up as well. This could be my long sessions with the laptop on which have caused it to die earlier than they expected it to, but honestly I would have thought it would have performed better than this.

The main problem is that I used my PC for skyping the US, particularly important as I’m going to be going home soon and need to arrange for an apartment in San Francisco as well as tickets, etc. Further it’s a big hassle considering that my exams start tomorrow though it has cut down considerably on the amount of time I spend sitting in front of the computer.

It will be a much bigger problem after exams are finished, and due to my rather cramped schedule, I’m not going to be in any particular place for the next month long enough to have it sent off for repairs. Fortunately, it is under extended warranty but it’s not doing me a lot of good.

I guess on the pro-side I do have access to computers still, it’s just not mine, and I will actually be going back to the US in a short time so if I can hold back long enough I’ll get it repaired pretty soon. Anyway, exams are coming tomorrow in less than 16 hours so I’m going to get back to work.