The Dummy Transistor Effect

Marketing technology is a dirty job. One pattern emerges over and over: hyping up the presence of  some technology whether it makes sense or not, as long as it sways some consumers.

Early transistor radios always prominently featured the number of transistors everywhere, and I've heard the tales of Japanese companies including extra ones just to up the count.  

I always thought that was just a legend, but a quick search turned up this amazing article from Billboard. In 1967 the FTC held hearings on the subject:

The clearest thing to come out of the Commission’s recent oral hearing on the “wild numbers rat race” in transistor advertising was that the whole issue is pretty cloudy. It also appeared that future developments in the complexities of engineering and design will make it even more so.

At the Oct. 4 oral hearing, Harry Gelbert, of Excello American Industries, a New York importer who sells to chain stores, department stores, retailers and wholesalers, said his company had to discontinue sales of “fraudulently” stamped transistor sets from a certain Japanese firm which stamped sets as having 14 transistors, when they had only six. Some had “dummy” and “tandem” transistors, some did not even bother to include dummies.

Gelbert said his retailer and wholesaler customers, deceived by numbers claims, complained when discounters would advertise a 14-transistor set for $2.95, when Gelbert’s firm had to charge $3.20 for a six-transistor set. A spokesman for EIA’s Japanese affiliate said the Japanese government had “censured” the offending firm, that standards were continually being raised on Japanese product, and he was sure there would be very little of this kind of trouble in the future.

But importer Gelbert said the "fraudulently" stamped sets were still pouring in “by the thousands” from Korea, Hong Kong and Okinawa.

Billboard, October 28 1967

I bet Mr. Gelbert sported wicked sideburns and the "censuring" by the Japanese government was most severe.

Another case in point is this amazing article from Wikipedia about the Jewel Inflation of 1960s.

It is doubtful whether adding jewels in addition to the ones listed above is really useful in a watch. It does not increase accuracy, since the only wheels which have an effect on the balance wheel, those in the going train, are already jeweled. Marine chronometers, the most accurate portable timepieces, often have only 7 jewels. Nor does jeweling additional wheel bearings increase the useful life of the movement; as mentioned above most of the other wheels do not get enough wear to need them.However, by the early 20th century watch movements had been standardized to the point that there was little difference between their mechanisms, besides quality of workmanship. So watch manufacturers made the number of jewels, one of the few metrics differentiating quality watches, a major advertising point, listing it prominently on the watch's face. Consumers, with little else to go on, learned to equate more jewels with more quality in a watch. Although initially this was a good measure of quality, it gave manufacturers an incentive to increase the jewel count. Around the 1960s this 'jewel craze' reached new heights, and manufacturers made watches with 41, 53, 75, or even 100 jewels. Most of these additional jewels were totally nonfunctional; they never contacted moving parts, and were included just to increase the jewel count. For example, the Waltham 100 jewel watch consisted of an ordinary 17 jewel movement, with 83 tiny pieces of ruby mounted around the automatic winding rotor


Bling bling

Adding gratuitous jewels to watches or transistors to radios might seem quaintly outdated, but I've encountered this technique in technology over and over in my career.

Once I was told of a project where the boss was enamored with a Microsoft product called BizTalk Server. I was never able to figure out what exactly it did back then (it appeared to be an XSLT transform engine of some kind), but it was marketed with gusto by Microsoft reps. It seems to exist today, and by jingo it has everything that you could want out of a piece of software: connect seamlessly, extend to the cloud, and go beyond integration. You wouldn't want a seamy connection, or extend a little lower than the cloud cover, or dammit, just simply reach integration and stop there. 

The Microsoft consultants who actually had to do the work used normal Microsoft products, like SQL Server and .NET to accomplish the tasks that they needed, but at the end they simply installed BizTalk and left it sitting there, not doing anything.

At the time I had to work with Microsoft SharePoint Server, an awful, awful monolithic conglomerate of awfulness. Developing native code for SharePoint was ridiculously difficult. But it had one component that everybody used to get out of having to write native code: an iframe that would let you write a little web app outside of Sharepoint and seamlessly bring it inside. Seamlessly, like BizTalk.

There's no limit to how much you can manjazzle your product these days. Cloud, Big Data, AI, Blockchain, Augmented Reality. I just think we need to start counting the clouds, and the datas, and say exactly how many crypto links are there in the block chains. The singularity is nearer and nearer with each day.

Elevator Wisdom Part 2: Elevators From First Principles or The Fractal of Underestimation

Behold this Twitter exchange. If you read my previous post, you can spot the surety of someone who thinks he knows something, but kinda does not.

Manual elevators do not have a throttle and a brake in the same sense a car does. They have what's known as a deadman, a spring-loaded lever that can move the car up or down at two speeds and stop is when returned to the middle position. It's not like driving a car or even a bike.

Secondly it is indeed true that the skill could be learned in a few hours. You can start your training in the morning, and by the end of the day do it well enough to work through a shift. But by no means does it mean that you can master this esoteric skill in a day. Mastering this skill takes years, decades, in fact.

Here's a typical elevator operator. Watch how smoothly he operates the controls, how he does not need to adjust the floor level several times, how practiced he is. To learn to smoothly stop level with the floor 99% of the time takes a lot of practice, and even when missing the mark, correcting smoothly is a hard skill to learn. You can stop higher, you can stop lower, you can miss the floor entirely. There are two speeds, you can switch to a lower speed too soon, or you can jolt the car by stopping too abruptly. Muscle memory does not develop in a few hours.

Beyond that, there are esoteric skills that are best not attempted by amateurs. For instance, the faster way to stop the elevator is to open the inner accordion door. You can time this maneuver with exact lining up off the floor and the car, and very elegantly increase the overall speed of operation without sacrificing the smoothness of the ride. A rougher, but more impressive maneuver involves stopping by reversing the lever — it is the fastest way to stop, but it supposedly damages the cables, although I saw oldtimers perform that trick from time to time. But most mind boggling is the consistency with which someone with 40 years of experience can stop the car exactly level every single time while keeping up the conversation and keeping an eye on the annunciator.

A skill seems complicated to someone who's ignorant of it, it seems too simple to someone with cursory knowledge, and complicated again to someone who has a good deal of experience with it. It's like some kind of a bike shedding fractal, I swear.

Elevator Wisdom Part 1: The Fear Ouroboros

"Futurama" Mars University (TV Episode 1999)

Ignorance is bliss, and knowledge brings fear. Or sometimes it's the other way around.

There are three levels of fearing or not-fearing something: out of ignorance, out of shallow knowledge, and out of deeper knowledge and experience.

I was once afraid of entering or exiting an elevator that did not come to a stop exactly level with the floor. If such a thing happened, it would stand to reason that the elevator was malfunctioning, and could possibly start moving once again when I'd be half-way in or out. My mind would paint gruesome pictures of being cut in half.

Elisha Graves Otis

Then, when I was in college, I worked as a manual elevator operator. I was in charge of a hulking Otis elevator that was built some time in the 1920s. In the course of my training I learned about Elisha Otis' ingenious safety mechanism.

This finely bearded gentleman invented an extremely simple mechanism. An elevator would have a spring-loaded device that would extend special claws into the notched guide rails if the elevator would fall too quickly. In addition to that, a safety switch turned off power to the motor when the doors were open. In the following years I was not afraid of being cut in half, and bravely dismounted my elevator in any position. 

Later on, working an office job in Midtown, I heard horrible news. An advertising executive tried to "catch" an already closing elevator by forcing her body inside. Instead of remaining motionless, the cab lifted her and squished her against the shaft, nearly cutting her body in half. The elevator remained jammed for several hours with two more people inside.

My first reaction was — this is impossible! The cutoff circuit, the safety device — what happened? What I learned, of course, was that technicians are able to override safety setting with jumper wires and put the elevator into a manual mode. Of course, sometimes they forget to remove the jumper wires, and an elevator can squish you as a bug if you try to force your way in. Now I am terrified of elevators once again, but for a different reason, of course.

Out of the three types of engineers,  Mort is fearful because he's ignorant, Elvis is overconfident because he thinks he knows things, and Einstein is fearful and paranoid because he has an idea about how the sausage is made.

47-50th Street Subway tower has an automatic routing system that promises to give you exactly what you punch into it.

In UNIX, I learned with power-greedy pleasure that you could kill a system right out from under yourself with a single command. This power was almost the first thing anyone teaches you not to do, then, with a devilish glee, tells you exactly how to: run as the user with complete systems permissions, go to the root level of the disk directory, then type in rm -rf *.

Ullman, Ellen. Life in Code: A Personal History of Technology (p. 43).

The second blog post in this series will focus on the very non-obvious learning curve of driving a manual elevator.