Is Good Documentation Just Talk?
Your Sept. 18 "Great Ideas Get Lost In The Sea Of Incomplete Documentation" \[p. 68\] is a subject dear to my heart. I'm resurrecting the design of a 25-year-old processor chip with penciled block diagrams, boolean equations, and implementation detailed logic diagrams to the Ada language. The IEEE HDLs are hard to use and read, and they're very expensive tools.
C was designed in the early '70s when short symbols were desirable for writability on clunking terminals. DoD had code readability requirements in place even before Ada language design. English identifiers, instead of cryptic symbols, facilitates self documentation. Most code style recommendations suggest generous comments. As the chair of a working group on documentation of a Jovial-Ada joint working group, I had suggested just the opposite: try to have readable codes without comments except when the raison d'etre isn't obvious.
EE Times did publish my letter in September on this language subset, but it raised no readers' response among several papers promoting the use of C to add "higher level abstractions." There's just no way to interest anybody to unify the design and programming of embedded systems with a simple readable language for reusable modules. Safety and security also are important for embedded systems, but most authors only talk about them.
The Economic Control Loop
I found your article "Economics For Engineers Is A Branch Of Psychology" \[Oct. 2, p. 160\] interesting. I had a letter to the editor in EE Times about 10 to 20 years ago with similar ideas—economics as a classic control system where loop delay and loop gain determined the oscillation frequency and lack of stability.
In my letter, I hadn't considered noise, but I don't know if that really affects stability. Introducing computers into the economic control loop does, as you indicate, decrease delay. But I don't think the reduced delay affects stability. Instead, it merely determines the oscillation frequency. In most control systems, reducing delay forces the oscillation frequency up so high that there isn't sufficient loop gain to sustain oscillation. Computers in the economic control loop, however, may actually increase loop gain. I think we've seen this with automatic stock trading. Though one investor has his computer set to a reasonable gain (buy this much if the stock drops this much), by the time you put all of the investors' computers together, the loop gain is very high, again leading to instability.
In my original letter, I used DRAM pricing and capacity as an example. Say there's a worldwide shortage of 1 million chips per year, and five companies each decide that they will fulfill that shortage. Seven years later, they have finished building their factories and are shipping the total of 5 million chips per year, creating a glut. Several of these factories shut down, creating a shortage, and the oscillating business cycle continues.... Interesting article! Thanks for the reminder!
Making Some Sense Of The Market
Before going to the point that led me to write you this small comment, I must congratulate you and the Electronic Design team for the excellent work that the magazine does every month. Congratulations!
I read Lwrence J. Kamm's article "Economics For Engineers Is A Branch Of Psychology." I'm not usually moved by these nontechnical subjects. But I always read his column with great enthusiasm, and I must congratulate him for your advice and comments. His columns are of particular importance for a beginner engineer like myself.
I started in physics, and then I jumped to the Euratom Fusion projects (where I'm still working toward my PhD) and made several control and data-acquisition systems. These were basically real-time systems for heavy ion beam injectors. In between, I also worked for National Instruments, Austin, Texas. I'm now working purely in software. And, sometimes it's very difficult for me to understand certain behaviors that the stock market shows constantly. In Europe, things are much quieter. Basically, all markets here follow Wall Street (with a six-hour delay). Lately, the tendency is clearly to follow not Wall Street exactly but Nasdaq in particular. But that's another story.
After I read Mr. Kamm's article, I remembered a book that I read once, The Crisis Of Global Capitalism, by George Soros. (I'm not certain that the book title is correct because I read the Portuguese version of it.) It gives a certain type of reply for your concerns in that article. I really must suggest that you read the book. I know you must be busy, but believe me, it's a rare book about economics that's worth the reading!
To start with, it's not boring. On the contrary. Mr. Soros' theory is that every "market agent," from people to enterprises, is ignorant about the market condition. So, like Kamm said, they make delayed and wrong decisions. But the most important thing about the book is its emphasis on the fact that every economic agent decision/action affects and changes the market. That is, every "control" action generates a different reply in the system! This isn't comparable to any physical system where there's a clearly defined transfer function and a determined and precise reply to a known input.
I think the man has a case there. Heck, he's very rich for some reason. In my small world of VME OS-9 real-time control systems, I simply try not to be irrationally exuberant.