Change of plans / Where are the challenges lurking? / Ethics, literature, and…

Change of plans

I went to the Media lab for the Psychology-class. 9 other students turned up too. But the professor were nowhere to be seen. We started to suspect, that not showing up on time is a clever trick to make us realize something about psychology. After 30 minutes, we gave up, one of us went and asked the department assistant whether the professor is going to show up or not. We found out that nobody knew where he was. So, we left. Later, Vineet and Uday told me that professor Ariely has taken 2 years off as sabbatical and is now doing something in Princeton. Apparently he was faster than MIT's bureaucracy. So, no psychology class for me in MIT.
What a disappointment this was! I just had to find another class.It should be preferable 9 units, convene once a week, in the late afternoon, not have a final nor midterm, have nothing to do with IT. Not a simple task to find such a course. It took me 1 hour to find the following:

11.540J Urban Transportation Planning.. History, policy, and politics of urban transportation. The role of the federal government and the "highway revolt." Public transit in the auto era. Analytic tools for transportation planning and policy analysis. The contribution of transportation to air pollution and climate change. Land use and transportation interactions. Bicycles, pedestrians, and traffic calming. Examples from the Boston area. Lectures: F. Salvucci, M. Murga

It must be fun. I am glad to be able to take a course by F. Salvucci, who is quite legendary figure in the field of urban transport. See here and here.
The class meets most of my requiements: it convenes only on Fridays from 2pm to 5pm, it has no final, it has nothing to do with IT. It is a 12-unit course, though, but all 4 assignment and the final project are done in groups and the groups seem to be quite large. So, I can lean a bit on the group. The course has some kind of semi-optional recitations on Monday or Tuesday evenings from 7.30pm to 8.30pm (biweekly or so). And then there is one walking tour around Boston. Sounds too good to be true!

Where are the challenges lurking?

The big money will not be in IT nor in biomedicine in the future. The big money will be in infrastructure. Especially, really big money will be in energy systems, in production of energy, transportation of energy, storage of energy and consumption of energy. The coming energy-booms is going to make the IT-bubble of the 1990s look insignificant in size. Therefore, everyone who has either feeling of duty (in making the world better or at least habitable for coming generations) or is just full on unsatisfied greed, will do well be studying some issues related to energy systems. Fortunately, developing the new energy system will be a task requiring exceptional system architects. And since most talented persons do not see this and concentrate in IT or bio, I will have no worries in the future.

Ethics, literature, and authority

I was able to participate in the Ethics etc course today. In the beginning, I was a bit shocked as I realized, that I was clearly the oldest student in the class. But then I realized, that it is just nice to have young classmates, who will certainly give me fresh ideas. And at least they are full of energy. The class itself was terrific. Of course, I was a bit confused how to behave, what kind of speech acts are expected by the professor and by the students. When I realized, that there is no specific format, I relaxed and started to participate lively in the discussion. The class is a bit too large for a real discussion, even though I must admit that professor Hafrey does a good job in keeping the discussion progressing. He is also able to remember, what everyone has said and to contrast opinions. After 45 minutes, I suddenly realized what the whole class is about. It is about understanding, that a) stories themselves can teach us a lot about ethics and authority and b) the ways we interpret the stories reveals a lot how we will interpret different situation in both personal and professional life and c) that stories also tell us quite a bit about us humans. I understood a) and c) before the class, but b) occurred to me just during the class. I think this class is really going to be useful.

Daily crono

Other things today. I took a walk in the morning. The weather is still nice: it was about 23C already at 7am. After hacking my thesis a little at home, I went for the non-existent lecture, and then used the 1.5 hours I save to print out and sort material for my thesis. Later in the evening, I went to listen to D.E. Shaw's recruiting presentations. The company seems to be interesting and they claim to have very meritocratic way of recruiting and also running the company. Maybe it could be a nice place to work for a while. Late in the evening, I read an interview with Fred Salvucci.
In other developments, today the boss of FEMA got finally fired. According to my memory, this was the fist time any of Bush's cronies got fired because of not handling his job. I hope, that this becomes a trend and finally the government will become better again. By the way, Paul Krugman claims, that the rampant cronyism and budget slashinghas driven many competent persons out of the government offices. If that is true, it is very sad and needs to be changed at once.

On complexity

I have now some 50 articles already, so it will take me at least 2 weeks to read them and make some notes and to get a reasonable understanding about the state of the art in software and system architecture complexity measurement.
It seems, that the area of system architecture complexity measurement has not been really studied at all. At least I cannot really find any really relevant literature. Most literature is about computational complexity (Kolmogorov, Chaitin, etc), which is relevant, but not really applicable to system architecture modeling as such. The reason is that knowing how expesive some model in terms of computation may not correlate to cost of designing, implementing, and maintaining the system, which the architecture defines. On the other hand, most of the existing software complexity measures, of which I know, require too detailed model of the software. So much even, that using them to estimate the complexity of SW before implementation is not usually that successfull. As an evidence, the Rapid Development-book by McDonnel (???) uses only lines of code in estimating the cost of designing and implementing a SW system (well, he also has different treatments of system, application, and shrink-wrapped SW) (I read some parts of the book yesterday at Kumar's place. I think I should buy the book, too)

There have been some efforts to apply existing SW complexity measures to measuring complexities of SW architectures, but so far I have found the efforts rather unsuccessful. Of course, counting the types and instances of object and processes (or, using the terms of Crawley: forms and functions) and interfaces between them will give some rather intuitive measure of complexity. Intuitively, the more there are forms and function, and the higher their inter-dependency, the more complex the architecture. Furthermore, one can give some intrinsic complexity values of forms, function, and maybe concepts used in sub-architectures. Using these as pragmatic ways to calibrate the measurement, one could maybe say quite much.

There are of course other possible approaches. One can consider the architecture as a graph and calculate some metrics of the graph. Or, one could apply the Kolmogorov-type approaches: maybe one build some logical formula in OPN, while running the simulation. Already now OPN keeps track of calculations by building complex formulas and simplyfying them only in the end Or so I have understood. So, if one would build a logical formula for each simulation branch, and then calculate some kind of Kolmogorov-measure for it, one could be able to say something "real" about the complexity. Maybe.
In general, I think that the possible ways to measure the complexity are of 3 types: syntactic, semantic, and judgmental. Calculating some metric of level of cyclicality of a graph is pure syntactic. Calculating some Kolmogorov-metric as I drafted above is more semantic, but does not need human interaction. Finally, one can assign complexity values to atomic parts or subsystems (manually, based on expert judgment) and somehow calculate a sum of it (as explained above). This is what I call (for now) judgmental.
But thesis is not only about developing some complexity measures. One (I) must also justify the whole effort of measurement. Of course, the intuitive idea is that a simpler (less complex) model is somehow preferable to a more complex model. This is what the famous KISS-heuristic is all about. And it is also, what in philosophy of science is called the Occam's razor. But even this is not as simple as it looks. There are several reasons: a simple model may not be extendable in the future: it is simple just because it is fixed, contains no options. Thus, maintaining and extending the model may become really expensive. Also, if we have several complexity measures, most likely they will not rank alternative architectures in the same order. What if we have 6 alternatives and 3 measures: then we can have all combinations of orders, and the whole exercise of measuring the models would have just increased the confusion. Based on these simple reflections, I assume that I really have to also discuss rather philosophical topics in my thesis. But that is fine: I am a certified philosopher.
Still, just having developed measures and justified KISS is not enough. I must also explain, what system architecting is, why it is important, and (to fulfill the management content requirement of SDM thesis) discuss what this whole shebang has to do with running product development organizations effectively. Furthermore, I need to explain what OPN is. But this I can well "copy" from Bill's excellent article, or at least find some inspiration there.
This is what I learned today while thinking about complexity.
I assume I am quite a bit excited about my thesis... the above text is more or less 3% of the text needed...and I am gathering some speed again as I can write this much blog ... I am a happy camper.

Alun perin julkaistu kuukausikoosteessa "Blogs from July to September 2005".