I feel comfortable relating this, having retired several years ago, and everyone involved has since moved on to other positions, but this remains one of the more frustrating episodes of my entire career. Not the most technically difficult. Not the most complex. Just the most infuriatingly unnecessary.

At the time, I was the lead sustaining engineer for a product that had been designed around the time I received my engineering degree. Yes, that old. Needless to say, maintaining a design with that much mileage on it was a challenge. Processes had changed, manufacturing specifications had changed, and component availability had changed, sometimes disappearing entirely and reappearing years later under a new name and triple the cost.

For context, about 30 years ago the government maintained a massive library of Mil-Specs covering everything imaginable. Many of these spelled out alloy compositions of metals and their allowable tolerances in exquisite detail. Eventually, someone decided that maintaining this library was expensive and that industry should take it over. Reasonable enough.

There was some minor pain for a few years while glitches were worked out, but eventually manufacturers were building to and certifying against commercial specifications. The old government specs faded quietly into history.

Except, of course, for our product, which had hundreds of drawings still calling out those long-abandoned government specifications.

Enter me. Or, more accurately, someone unlucky enough to be me. We had to comb through old specs and new ones and decide whether they were close enough to be considered equivalent. If they were, the drawing had to be updated. If they weren’t, testing had to be performed to justify the change. Slow, tedious, unglamorous engineering—the kind that actually keeps products buildable.

Once that was done, we had to notify our customer that a change was being made. Changes came in two flavors.
Class I: a major change affecting the product’s characteristics, form, fit, function, reliability, safety, and a dozen other things people argue about in meetings.
Class II: a change that did none of that.

Class II changes bumped the revision letter. Class I changes were a whole different animal. The product had to be fully traceable in case something went sideways years later. That usually meant part number changes, and on a configuration-controlled product, that meant pain. Expensive, bureaucratic, soul-crushing pain.

When we submitted a change, we classified it as Class I or Class II and attached whatever data and test results justified that decision. The customer had 30 days to review a Class II change before implementation. If they disagreed with our classification, we got to do more work and resubmit it, joy.

On the customer’s side, my counterpart would review the package and decide whether it was truly Class I or II. If it was Class I, a configuration control board had to be convened, multiple engineering disciplines, managers, and people whose calendars were always mysteriously full. Class I evaluations were expensive, time-consuming, and best avoided unless absolutely necessary.

All of that brings us to the absurdity.

We were required to change a stamped aluminum label to a computer-generated Mylar label. Not a new concept. We had been using identical Mylar labels on other products shipped to the same customer for years without incident. The label was roughly 1 x 2 inches, legibility improved, and nothing about the product’s performance changed in any measurable way. As a Class II change, it was about as obvious as gravity.

Then configuration management got involved.

After reviewing the change, CM decided that the difference in weight between the two labels made this a Class I change.

Let that sink in.

This label was going on a box weighing about 5 pounds. The aluminum label weighed roughly 0.09 ounces. The Mylar label weighed about 0.0032 ounces. That’s a difference of 0.087 ounces on an 80-ounce product, roughly 0.11%. Slightly over one one-thousandth of the total weight.

Yes, there are scenarios where that matters. A spinning hard drive platter, for example, would absolutely care. This product, however, contained no spinning parts whatsoever. It was a box full of electronics. The weight variation of the screws inside the enclosure exceeded the difference between the two labels.

I explained this. Calmly at first. Then less calmly. Then with diagrams, numbers, and what I believed to be unassailable logic. My supervisor, having worked with me for years and recognizing the early warning signs, escalated the issue to the director level. I’m fairly certain he was trying to head off a phone call from HR.

To be clear, I was only months from retirement. I could have shrugged and let it go. It would cost the customer some time and money to reject our Class I change and tell us to resubmit it as Class II. Embarrassing, sure, but survivable.

But pride crept in. Not just for myself but for the engineering department I had help build.

I did not want to establish a precedent where configuration management overruled engineering judgment on engineering matters. I had worked hard to build trust and mutual respect between the engineering teams at both companies. Now we were about to look like idiots—submitting a change as Class I because someone, somewhere, decided that math was optional.

You can’t win them all.

The change was submitted as Class I. The customer rejected it almost immediately and asked, politely but pointedly, what the heck we were thinking, because it was obviously a Class II change. Shortly thereafter, I retired.

Between this and subsequent changes submitted as Class I for the same reason, our CM department’s arrogance likely cost the two companies well over a hundred thousand dollars. Worse, it eroded the confidence our customer had in our engineering department, leading to endless delays as the customer demanded more data, more tests, and more justification for changes that should have been routine.

Writing this now, I can’t help but feel a sense of camaraderie with the professionals who once worked at the FDA. Decades of training, research, and expertise dismissed, not because they were wrong, but because they relied on data instead of fear, noise, and people who felt strongly about things they didn’t fully understand.

 

 

---

---

And of course, today’s song from Songer… Dance With The Data

© 2025, Byron Seastrunk. All rights reserved.