Powered by glolg
Display Preferences Most Recent Entries Chatterbox Blog Links Site Statistics Category Tags About Me, Myself and Gilbert XML RSS Feed
Sunday, Mar 25, 2007 - 17:50 SGT
Posted By: Gilbert

- - -
Projecting Ire

About three weeks remaining for three computing projects, which I'm getting a bit hung up on. The big question would be: How much work is expected for an A? Oh, and to those who say "just do your best" or "it's not about the grades, it's about the learning process", I thankyouverymuch in advance, ah.

I first got hung up on this when I attempted to build a working mini-desalination plant for my Social Studies project in primary school, with the individual structures made out of aluminum-foil wrapped mounting board fronted with the transparent plastic sheeting used to protect schoolbooks. The pipes between the structures were similarly manufactured, and the stages of desalination lifted from the textbook, with the whole thing mounted on a styrofoam board. There was one little problem - it leaked early on, and obviously didn't work. And it scored less than a map of Singapore made out of different coloured beans.

That was not the last word on harebrained ideas, of course. My idea of producing heat with a bunch of shortcircuited AA batteries to melt ice cubes was soundly beaten by another team that simply crushed the ice cubes and mashed them in their hands, but that I have no complaints about.

More than ten years on, it is more of the same in university. I get an idea that appears new, useful and good, but after I state it, it looks more and more to be rather trivial, the sort that anyone beginning from the same starting point would have come up with. For example, given a problem that requires some form of search, there are only so many search algorithms that are known, and it is hardly an exercise in imagination to try all of them. Then, given a composite problem, there remains only the question of finding all the ways into which it can be broken up into subproblems, and then for each of these subproblems, try all the methods that may work.

Mere enumeration and testing obviously will take time, and the difference between great researchers and those of the ordinary ken would probably be (A) knowing more methods and algorithms, (B) being able to apply them quickly, and (C) being able to infer relationships between subproblems, and mentally eliminate those methods that will not work, without trying them (fellow CS students may recognize this as pruning). Even this much is not obvious, though, as evidenced by the esteemed Hungarian mathematician George Pólya, extremely influential 1945 classic on problem solving, How to Solve It.

Its brief outline is as follows:

1. First, you have to understand the problem.
2. After understanding, then make a plan.
3. Carry out the plan.
4. Look back on your work. How could it be better?

Very reasonable, but even the clever among us sometimes fall prey to not following this framework. Recently, I heard a fellow student asking a TA why his program wouldn't work on his system. The TA asked him whether he modified his program from the example program given. The student replied that he had. Then the TA asked him whether that example program worked on his system. The student replied that it didn't.

Duh. But everyone has their off-days.

Well, Edison said that "Genius is one percent inspiration and 99 percent perspiration", and I hope he's right. Time to just sweat it out and hope for the best.

At least two of those projects are individual work, which simplifies matters. Fred Brooks' famous "Bible of Software Engineering" (so named because everyone reads it but nobody actually practices what is being preached in it) The Mythical Man-Month states that "adding people to a late software project makes it later", also known as Brooks' Law. I had previously accused Brooks of just stating the obvious; In another of his famous papers, he proposes approaching essentially difficult software engineering problems by training great designers - imagine a scientist concluding his paper on a physical problem by suggesting that we increase educational funding for future scientists! However, on reflection, that was one of the few honest answers that could be given to such a wide problem, and in any case this world needs more common sense, not less.

The title of The Mythical Man-Month arises from the observation that software engineering (and indeed many other jobs) cannot be hastened by throwing more men on the job. To illustrate, say we want a one kilometre long ditch. One man might be able to get the job done in ten days, five men in two days, and ten men in a day. The man-month is a valid unit of work in this case. However, if we want a one kilometre deep ditch, the extra men will not be able to work at a deeper depth simultaneously - the time taken then will not be reduced by any significant amount even with an army to do the job.

Each project probably has an optimal number of people that should work on it. Say if there is a large number of simple individual projects, it may be best for the group members to divide them and work on a number of them individually, rather than attempt to collaborate on each one. Having too many people to begin with will often degenerate into politicking - the more conscientious will worry about not doing their fair share, and the lazy will worry about not appearing to do their fair share.

There is probably nothing so sad (and fatal) as a conscientious but incompetent group member, though. One knows that he is trying his best, but is simply not up to it, whatever he does will likely have to be greatly edited or fully redone, which may be taken very badly by him. From a larger perspective, it might be most efficient to tell him to just sit on the sidelines and twiddle his thumbs - I am reminded of the Dilbert cartoon where a rather hopeless employee boasted that he worked overtime on the job, and the rest responded that they all had to work overtime the next day to fix the mistakes he introduced. Yes, there is always the soft option of just letting him "do his part" and submit the project with huge flaws, resulting in Bs or worse for everyone. Truly a tough decision. At least the irresponsible and incompetent group member will relinquish his task with few protests.

The corollary from this is not to be too insulted when not assigned a significant part of a project. It is highly possible that the project was really too small to be evenly distributed anyway, and the single significant part, if there is one, is best tackled individually.

Moving on to local news: Our worthy ministers' salaries (currently about US$800k a year) are set to rise, "to close a 55 percent gap with private sector income levels".

Let's begin by considering how ministerial pay is determined. It is supposed to be pegged to "two-thirds of the median income of the top eight earners of six professions". Not too sure about exactly what this means - do we take the median of the 48 rich guys, or take the median of the six professions and average them? In any case, it's a lot of money.

A few observations: One, is private sector pay consistent? It is not improbable that pay fluctuates greatly on an annual basis - a lawyer may win a lot of big cases one year, but not be as fortunate the next. On average a small number of them will have great years, of course, and guess who contributes to the peg?

Two, the median pay of the top eight or 48 is hardly reflective of the well-being of society at large. If this median pay suddenly shoots up, does it mean that the ministers have done well? Probably fairer is a peg to say 100 times the salary of the median Singaporean, although on second thought that may not sound very palatable. Oh, and heaven forbid our search for talent attracts Bill Gates to Singapore.

Three, the private sector is likely far more unforgiving than the civil service. Fail to meet expectations in a competitive environment, and one's out of a job. It is an open question if top civil servants would actually be able to secure a comparable salary, for years on end, in the outside world. One response to this is that ministers do have a competitive check - they have to face elections every five years. Please excuse me while I roll on the floor.

Four, the main justification of high salaries has always been that if salaries are low, top civil servants would either jump to a higher-paying job, or turn corrupt. I don't know, but if I walked up to them and asked if they would be tempted to practice corruption if they were paid the meager income of say S$20k a month (about a fifth of their current income), I would be entitled to expect no for an answer. But, part of Singapore's operating philosophy has always been pragmatism, after all. Here, service to the nation coincides with service to the self, and it is impolite to enquire about the two separately.

Five, the salaries are tellingly out of proportion to the rest of the world. For leading the world's greatest superpower, the President of the United States receives US$400k a year, less than half what our Prime Minister gets, and also less than what a normal Minister gets. Look around the rest of the developed world, which we have attempted to emulate in many respects - nowhere else do we see such figures. Surely they face the same issues of brain drain and possible shenanigans? Perhaps they do indeed supplement their income through other means, in which case we can rightly be proud of pre-emptively supplementing income through official means. Singapore rules!



comments (0) - email - share - print - direct link
trackbacks (0) - trackback url


Next: Cold Shorts


Related Posts:
Open Bookame
On Economics
Economics Thus Far
Taxing Two Ways
On The Money
Back to top

Copyright © 2006-2025 GLYS. All Rights Reserved.