I just read your April 1 column on "What's All This Doctor Science Stuff, Anyhow?" You don't know how much that humorous article brightened my day. (Of course you don't. It's metaphysically impossible). Several years ago, while doing a contract job in Norfolk, Virginia, I was listening to the local rock station and they did a series of "commercials" for the Heiney Winery (run by Big Red and his brother Thor).
Upon investigating, I found that it was a syndicated script that stations could subscribe to and produce themselves. I thought it was really clever, as is so with Doctor Science. I'm going to find out who airs it around here. If no one does, I'll try to lobby for it.
You will no doubt get mail from many readers decrying that program as silly (it is), outrageous (yes), unbecoming for professionals like us (darn tootin'), and incomprehensible (not for people with a sense of humor). But you know what? I think it's really neat. Humor, especially the wry, scientific type, requires a very high intelligence. The fact that we appreciate it makes us better than everyone else (or at least certainly smarter)!
So bravo to you. Whenever we start to take ourselves (or our jobs) too seriously, we should think of Doctor Science. I'm going to compile a list of electronics hoaxes for your next April Fool's column and send them to you.
The good Doctor sure keeps us on our toes! Unfortunately, KCSM has dropped his program, so I now subscribe to his tapes from Duck's Breath (1-800-989-DUCK).—RAP
I have also been keeping a list of backpacking equipment. I started in 1987. In 1990, I changed from a simple list to a spreadsheet so I could watch the ounces accumulate into pounds. It's more productive to worry and fret over weight at the keyboard than it is at the trailhead. It must be a common trait among engineers to measure things and make lists. My wife refers to this as a fault, but she was an art major and must be forgiven for some of her views.
Your data indicate that the sum of all the weights from the spreadsheet is less than the pack's weight. Which do you believe?—RAP
I'm on your side in the digital dispute. I complain about overflow or roundoff and am told, "that's not error; that's just the way it works!" Well, according to my Funky Wagnalls, it's error. Would these people be content with jagged lines and purple people on their TV set because "That's the way it works?"
When I get something that's obviously totally wrong I'm told, "It's not computer error, it's just a software problem!" That's little consolation when the computer turns your engine off in traffic. Would these people accept a paycheck that's wrong (on the low side)?
I am told a circuit has to work because the computer simulation says it will. In this case, fortunately, I can breadboard the circuit and it doesn't work. Furthermore, breadboarding takes less times, and when I am done I have a real circuit. Now if the circuit had a million transistors, I would have to simulate it.
What about when I go to the parts place after calling first, and the guy comes up empty-handed. "The computer said we had 10, but the bin is empty." I'm sure somebody punched in the wrong data, or no data at all, but it's cost me my time and cost them a customer if I have an alternative.
And how about the guy who was executed by software? He complained that he was getting too much radiation, but the computer said no, and of course the operators believed the computer. I'm sure it was a "small problem," except to the patient in question.
I am thankful for my digital calculator when I need more than a few digits of accuracy, but I have found that if I accidentally divide by zero, the only way to ever get it to do something useful again is to unplug it and start over. On the other hand, my analog divider recovers immediately.
My digital scope has great memory and hard copy, but it often displays waveforms totally different from what's really happening. Of course, when I finally figure out the real waveform, I can get a good picture. But if I know the waveform, I don't need a scope.
I keep an analog scope connected, too—the combination works great. My digital spectrum analyzer is many times better than the old analog job, but neither is always truthful.
I love my synthesized function generator with its five digits, but it quits without warning if you set the first two to zero. I have a cute digital clock that is far more accurate than my analog watch, unless you shuffle across the carpet in dry weather to pick it up, in which case it becomes a random number generator. On the other hand, the analog watch does lose a fraction of a second if you bang it hard.
Something is wrong when one gets a reputation for being "anti-digital" just by pointing out known facts. I think it's basically a question of intellectual honesty, which leads to proper use. Computers are powerful, but when they are taken to be infallible we are in for a heap of trouble.
Safeguards, crosschecks, and over-rides don't get built in. Mistakes don't get corrected because "It's in the computer." People believe data that should be questioned.
Computers are bad when they make you think you can do something that you can't. The same goes for PhDs (I have one). And professional societies. Some computer advocates are honest about their product and some are right down there with used car salesmen.
Mt. Airy, Maryland
Art, you're the most skeptical so-and-so I've seen since I looked in the mirror this morning. Great! Now, can I sell you some Fuzzy Logic?—RAP
Dear Mr. Pease:
Your battery anecdotes reminded me of the new 3.3-V logic standards for computers that operate on batteries and at high speed.
The earliest popular vacuum tubes were the 2-30 and 2-33 with 2-V filaments designed to suit lead-acid storage batteries. Very few people in those days could afford a 2-V battery in a glass jar, plus the charging equipment (Tungar). Instead, two A-size dry cells were commonly used—each about the size of a small thermos—plus a rheostat to drop the voltage back to 2 V.
The Philips 109A tube had a 1.3-V filament (good with a half-spent battery), with about half the current of the 2-30 tube. It ran on a very low plate voltage (I forget how low) and was very economical to operate. The U.S. tube industry never got around to designing tubes to fit widely available batteries. The industry initially picked a 2.5-V filament for ac (before the indirectly-heated cathode). This eventually gave way to the 6.3-V standard that ushered in automobile radios.
Now 60 years later, I see a sea of 3.3-V IC designs intended for battery operation, but I do not see popular battery types that fit this voltage. Hence, we have a repeat of standards sans rationale.
It was recognized at the dawn of computers that to get higher speeds, one must limit the logic voltage swing. In 1955, I was responsible for a 36-bit machine that used germanium DCTL circuits; the arithmetic unit operated at 3 V. The swing was between about 0.01 V (saturated collector) and 0.3 V (base load).
Sometimes not knowing technical difficulties prompts bold (or rash) development decisions. Because it was a solid-state machine in 1955, we needed a memory-access monitor. A CRT would have worked fine. However, a research manager approached me and said he could make an electrolumines cent display for $1000, and the next one for $150. I ordered two made. The display was delivered and even worked when two lines on the 64-by-64 matrix were excited. And it could even be seen in total darkness with dark-adapted eyes. We also figured that we could not make reasonable-solid-state drivers for it, so we finally used a CRT.
Incidentally, that machine was codenamed Solo by the funding agency. It was intended to be operated by a single person and was housed inside a specially-designed desk. Could I qualify as being the first to make a personal computer—at least a solid-state one?
1955? That's before I built my first Heathkit, so I can't comment on your claim. But you're probably right. All portable computers do rely on switchers for best battery usage. —RAP
Yet more on spreadsheets. Duncan Hall and myself developed HandiCalc—a virtual spreadsheet—shortly after VisiCalc arrived as a response to the rise of blind faith in spreadsheets that we were witnessing. Unwilling to emulate pencil and paper, HandiCalc consisted of a pencil and paper (gridded engineering calculation sheets) and was "powered" by a between-the-ears wet processor. Strangely enough, when coupled with an HP 12C coprocessor, HandiCalc outperformed our colleagues' Lotus 1-2-3 NPV calculations, and it was accurate.
Your observation on misused spreadsheets is in fact a special case of Prof. Arthur Williamson's (University of Canterbury) "Reality Failure." "Reality Failure" is the general case where reality fails to comply with the rules and assumptions imposed upon it by a design engineer. Another special case of "Reality Failure,"—well known to Free-marketeers and Economists—is "Market Failure."
Christchurch, New Zealand
Better your processor than some of the "wet-behind-the-ears" computer guys. And thanks for pointing out the "Reality Failure" situation. —RAP