Wednesday, August 27, 2008


其實我不認同社民連的某些立場,特別是經濟方面,只是我覺得現在的傳媒全都很「乖」,對社民連只有負面報導,令佢地淪落到只有在那個冇人聽的網上電台同埋去榕樹頭講故... 好似呢D咁大快人心嘅場面,如果唔係因為選舉要公平分配時間,邊有電視台會播?其他那些民主派,個個咁錫身,邊會咁鬧?

Tuesday, August 19, 2008

degree inflation

I think I should keep my mouth shut, so here are just some links:

Watchdog: degree grades arbitrary

Email leak of degree inflation

Or maybe I could say just a little...

The one reason that I've seen to support the current degree classification instead of GPA is: "students' effort and qualities cannot be represented by a single number". Well, if it cannot be represented by a number then why can it be represented by a even more meaningless label such as "upper second"? (Those labels remind me of the credit crunch crisis: there are only two quality levels for mortgages, prime and sub-prime. They may sound like "very good" and "not as good", but actually they mean "acceptable" and "rubbish".)

If it cannot be represented by a single number then give me an array of numbers then. As someone involved in admissions, I like neither GPA nor degree grades. Just give me all the marks and the average.

As for "arbitrary grades", the following is a purely hypothetical situation:
[1 hour 45 minutes into a meeting]
A: Now we have a Mr. X who got this set of marks, which is just a bit below 2-1 degree, should we consider putting him to 2-1?
B: But according to our regulations, rule 37.1.2(a)(iii) says he needs at least N marks in M subjects in order to be considered as a borderline case.
A: Yes but if we raise this mark by 0.5 then his marks satisfy the condition in rule 4.69.12(f)(iv), after we execute this rule then his marks satisfy rule 15.1.2(b)(ii), after executing this rule then his marks satisfy rule 37.1.2(a)(iii) which means he can get a 2-1!
Of course, it certainly isn't like this in reality, is it?

Sunday, August 17, 2008

A trivial algorithm

I seldom talk about my research here. First, there's not much (...); and second, most of them are on very simple (a.k.a. "fundamental") problems. Recently, we have a paper where the "main" result is so simple that some of the reviewers' comments were:
  • simple to the point of triviality
  • The new algorithm is almost trivial. It can be seen as a "real" result only due to the fact that several researchers tried to design improved algorithms in the past but did not succeed.
(Actually, I plan to put these comments on the slides in the talk.)

So here is the problem. A set of equal-length intervals arrive online. Each interval carries a weight. The objective is to choose a set of non-overlapping intervals in an online manner such that the total weight of the chosen intervals is maximized. An interval can be discarded (preempted) in favour of another newly-arriving interval.

A deterministic 4-competitive algorithm is known and is optimal (Woeginger, TCS, 1994). The question is: give a better randomized algorithm.

Our trivial algorithm is 2-competitive. Since it is so trivial I'll not describe it here but leave it for you to imagine. Either people do not care about the problem (except this, this and this) or people have come up with this many times and didn't bother to write about it...

Friday, August 08, 2008

cosplay @ leicester...

我仲以為香港先有「動漫節」呢味嘢, 點知今日係學校出現大批 cosplay 人物... 原來本校歷年嚟都主辦呢項「盛事」, 仲要係全UK最大 (第二樣又唔見咁叻...), 只係通常每年呢個時候我都唔係到, 所以先唔知. 我一行入 canteen, 見到一大班鬼仔鬼妹係到扮架仔架妹, 感覺很 surreal...

Thursday, August 07, 2008

What should I post?

I originally hoped that this blog can contain posts on a variety of topics - academia, politics, fun, teaching or even research. Over time however it becomes more or less a place of complaining about my miserable life, mostly about travelling. I hope to change this, and to make sure I do, here is a list of topics that I will post in the near future:

- how algorithms teaching went here
- project supervision and project reports
- politics: upcoming Legco'08 election

I'm not so sure how I can write about the teaching thing though, since there are many things like this that I can write but I don't want to get fired just yet.

Also, in order to update this blog more frequently without putting in any content myself, i would just quote other people's stuff that I've come across.

Here is today's quote:

(倪匡, apple daily 07/08/2008)