|
book reviews
memoirs of scientists and engineersreviewed by T. Nelson |
|
|
book reviews
memoirs of scientists and engineersreviewed by T. Nelson |
Reviewed by T. Nelson
Richard Hamming, who died in 1998, is one of those names that most engineers and scientists are familiar with. Hamming codes and coding theory are taught in every course on communication theory along with Shannon’s famous theorem. But how did Hamming succeed? Hamming says that every scientist and engineer must have a vision. If ultimate success is having your name used in lower case, as Volta, Ampere, and Coulomb achieved, you need a purpose in your career. Getting anywhere is better than drifting and following the crowd passively as most people do.
Some readers might disagree about the names: we also have hitlerian, machiavellian, and satanic. Hamming would have needed to define a unit to gain lower-case status. “Earl, try increasing the filtering by three or four hammings” just sounds funny.
Hamming believed that ‘learning to learn’ can’t be taught by abstractions, only anecdotes. So the first half of the book is filled with anecdotes and lifestyle advice: Question everything everybody says. Question your own beliefs. Do back-of-the-envelope calculations (now known as BOTECs) and napkin calculations to make sure an idea makes sense. Inspect all definitions to find out why they were made. Prepare for greatness by learning everything you may need. Luck favors the prepared mind.
His favorite anecdote is that the path followed by a drunk follows what mathematicians call a ‘random walk,’ where each step takes him on average √n steps from the origin. But if he sees a pretty female the drunk changes to a direct path, where each step moves him a distance of n. That is what you need to do: focus on the ultimate goal and learn whatever you may need. It will give you a √n-fold advantage.
Okay, but how do you know what you may need when everything you learn will be obsolete within twenty years? Doesn’t everybody do BOTECs and napkin calculations all the time at lunch? And don’t scientists already have such enormous egos they all think they’re already great? Maybe, but most don't see the big picture and they end up studying minutiae.
Anyway, nobody really uses napkins. In some places the napkins are cloth and if you write on them the maître d’ will just glare at you, while in others they’re paper and they get smeared by grease. So a true great always keeps a piece of paper in his pocket protector.
Hamming was a mathematician, so he derives life lessons from math. When the math for filling a cube with spheres gives you the wrong answers, you can either ignore the math as most people do or dive into it as he did to find out what went wrong.
In the second half of the book Hamming uses coding theory to make the anecdotes more specific. When Hamming noticed that many engineers were being sidelined at Bell Labs because they stuck with what they knew—analog circuits—Hamming learned all he could about digital stuff and collaborated on a book with an expert in the field. After finishing his half he discovered that his co-author hadn’t written a single word, so he ended up with Coding and Information Theory (1980) by R.W. Hamming without the other guy, who will forever be known to history as ‘whats-his-name.’
(Actually it was James F. Kaiser, who wrote or co-edited eight books of his own. Kaiser might have been following the maxim “You can accomplish anything if you don’t care who gets the credit.” Without him the book wouldn’t have been written. So maybe that makes him even smarter than Hamming.)
Another Hamming tip: if you think something is impossible, write down the reasons why so you can change your opinion when necessary. (That makes six hammings so far.)
Maybe writing a scientific book as a form of self-education is overkill (though it worked for Steven Weinberg), but as Hamming says, however you learn you must do it continually. You must learn stuff. And so we get a bunch of stuff on coding and information theory to learn.
Even here there are questions. Why do we use Fourier series instead of something else? Why is the Nyquist limit so important? Hamming uses his BOTECs, in this case testing a simple differential equation (y′′ + y = 0), to get answers.
About Fourier functions he says the complex exponentials are eigenfunctions of the sampled system. The famous transfer function is nothing more than the eigenvalues of the corresponding eigenfunctions. What ‘eigenfunction’ means isn’t important. What’s important is satisfying one’s curiosity instead of ignoring it. One might say you can trust your brain to tell you when something is important. The brain knows what it needs, though it might not tell you why. All you need is to do what it says.
Suppose we ask about AI. Is what they’re calling AI really AI? Is it really new at all? In Chapter 7 he writes:
[A]rtificial intelligence is not a subject you can afford to ignore; your attitude will put you in the front or the rear of the applications of machines in your field, but also may lead you into a really great fiasco!
This was written in 1996 and it’s still true. In chapter 15 (Digital Filters - II), he expands on the idea:
[P]eople always want to think that something new is just like the past—they like to be comfortable in their minds as well as their bodies—and hence they prevent themselves from making any significant contribution to the new field being created under their noses. . . .
When something is claimed to be new, do not be too hasty to think it is just the past slightly improved—it may be a great opportunity for you to do significant things. But then again, it may be nothing new.
Hamming was wary about computer simulations, especially in the soft sciences:
I suggest you keep your integrity and do not allow yourself to be used for other people’s propaganda; you need to be wary when agreeing to do a simulation!
. . .
[B]eware of any simulation of a situation which allows the human to use the output to alter their behavior patterns for their own benefit, since they will do so whenever they can.
In another anecdote, Albert A. Michelson (made famous by the Michelson-Morley experiment) noticed that his machine to calculate Fourier terms gave an overshoot. Every mathematician but one told him his equipment must be faulty. But when J. Willard Gibbs (known for Gibbs free energy) looked into the question he rediscovered a forgotten fact that is now now known as the Gibbs phenomenon. Gibbs is now famous. Nobody knows who the other guys are.
What’s true for digital filters, which replaced analog ones, might be true for LLMs. Or it might not. As with everything else, the only way to decide is to understand.
may 19 2026