Size matters

Sept. 29, 2014 – It’s been proven time and time again — small groups perform better than large groups.

Lockheed’s Skunk Works projects, for example, were created so that planes could be designed with little bureaucracy. And over the years, this department has produced a number of famous designs — proof that the fewer cooks there are in the kitchen, the more productive you can be.

Writing this column reminds me of an experience I had indirectly with Lockheed Skunk Works. Years ago I met an engineer for lunch for another reason, and it turned out he was one of the senior people at Skunk Works. He had seen an interesting truck go down the road and decided to follow it to its destination — Chrysler, which was adopting the skunk works model to their products. As a car nut, I couldn’t resist. I said, “Show me,” and he immediately arranged a visit.

We went to the Chrysler research facility and checked out the section where they were designing new trucks. A small group, in fact, was designing trucks, and their latest design was posted on a board. It was beautiful. I said, “It looks like a Peterbilt.” The designers were very pleased. They took a different approach. Instead of taking a car and upgrading it to a truck, they took the highway monster and downsized it to a pickup. In other words, they came from the other direction. To this day, the design shows in the truck — the Dodge RAM. (My noble secretary, Debbie, is on her fourth RAM truck.)

We need to take the bull by the horns and really believe that skunk works is the best way to unleash a new product.

Likewise, successful startup companies typically employ three to seven people. We also know that boards of directors should consist of an odd number and be about five in size to be productive.

There is a book that you all should read (or re-read) called The Mythical Man-Month. I’ve talked about it before in my column. The book points out that if you have a software group with seven people and add one more, you increase cost and time of delivery. More people does not equal more productivity. My experience with startups suggests that three to five people are the optimum human pack size.

One of my talks at the 2013 ISA Marketing and Sales Summit discussed pack size in some detail. The actual experiment was done at the COFES conference hosted by Cyon. We did not plan the experiment; we used serendipity.

But first, I should define a pack. A wolf pack is not one wolf and not 100. One wolf cannot bring down game, and if a pack of 100 exists, the game you do get is not big enough to feed the entire pack. So the wolf pack settles into a size that matches the environment. Even geese form packs. My daughter raises geese on her farm, and if she injects a new gander into the group, turbulence occurs. As a teenager, I was involved in social groups on Long Island. We all belonged to a “pack” (gang) consisting of three to 10 kids. It provided us with survival and safety.

Now back to the conference I mentioned earlier. We actually measured the human size of a pack by accident. The host at the COFES conference, Brad Holtz, wanted some after-dinner entertainment on the future of software, so he asked me to come up and entertain. We decided to do an IQ test. We had 300 people at 30 tables — 10 per table. They did not want a morally physical test on quantum mechanics, but rather one that most software engineers and automation experts could solve. We estimated that one-third of the engineers at the dinner would be able to solve the test. What was it? We asked attendees to complete the following sequences:

2, 5, 10, 17, 34 …
18, 28, 45, 90 …
Solid, liquid, gas …
How does the earth stay hot?

No table could solve the whole series. Although three engineers at each table had the right answer, they did not manage to give a group answer. This says that the pack size of the experiment (10) is too large to be effective for humans. What does this mean? Sell to nothing larger than a pack (a single entity). Larger collections cannot decide.

P.S. The answer most could not get was: solid, liquid, gas … plasma.

This column originally appeared in the September 2014 issue of Manufacturing AUTOMATION.