Often jobs are created for purposes that aren’t readily apparent. It might be as simple as creating a job so the boss’s daughter has an opportunity, or so the company can comply with specific regulations. Understanding the true functions of jobs is critical to navigating the corporate maze.

One of my internships in college was at a large natural gas provider in Boston, and I started with great excitement. The new job would actually use my statistical and economic education! This was the big leagues! Or so I thought…

The first day, my boss, a brilliant thirty-something MIT MBA graduate, showed me the economic model he had built. It was a massive, complex model taking in everything from population numbers to projected growth statistics, even long-range weather forecasts. All to produce a single output—the projected percentage increase in the amount of natural gas that would be used in the coming year.

My first task was to go down to Boston City Hall and enter employment data into a laptop from large books they kept there. That data was broken down by region and industry. In some cases, I had to cajole the librarian into giving me access to special books in the back room. This was data entry, pure and simple.

The promise was that after a month spent entering the data, I would be able to help with analyzing it. So, after a grueling month bent over a laptop in a dimly lit room, I finished the data entry and started on my glorious programming career.

My first job was simply to import the data I’d collected into the large mainframe that was running the econometric model using the PL/1 programming language. That was no small task, given the unfamiliar programming language and machine. Still, it was the sort of work I loved, and I did well.

In fact, I did such a good job that they asked me to program more and more pieces. So, over the next few months, I worked hard to program logistic regressions, graphs of employment trends, and endless long-term weather predictions. Well… more long-term weather guesses. Best of all, what I was doing was critical to the company. The data would allow the company to understand when and where natural gas would be needed so they could make the appropriate investments.

After many months, it came time to run the model from beginning to end. I had tested every component, and I knew that it was doing exactly what it was designed to do.

With my boss standing next to me, I ran the complete model for the first time. It came out with the result, a forecasted 2.2% increase in gas needs for the next year.

“I think it should be more like 3.5%,” said my boss. “Check it over and see what’s wrong.”

But I had checked each and every piece rigorously. In fact, I’d kept print outs of the tests. I got them out and went through them piece by piece to show my boss that the model was, in fact, running exactly as expected.

“Well, I still think it should be about 3.5%. Figure out what numbers you can tweak to get it there.”

I was aghast. I had really believed in my data forecast. In class, we were taught to trust the models we build and follow their guidance. But here, my boss was telling me to not only ignore the output but to change it. As a staunch follower of the scientific method, to me, this was sacrilege. Instead of following the evidence, we were predetermining the outcome! Why go through all that work to pull this together only to ignore it!

So, I confronted him on it. If we believed in the model and then played with the numbers, the company would be building too much capacity next year. It could be disastrous.

“Troy,” my boss said, “you don’t seem to understand. This data isn’t used to run the company. No investments will be made from it.

“You see, we are a highly regulated industry, and we’ve found that if we have a model that is really complex and detailed, then the regulators will allow us to raise our rates more.  The regulators don’t really understand these models, so we make sure they are complex enough to impress them. We make sure they contain variables that seem predictive, like employment and population and weather, but we don’t rely on this to run our business.

“In short, it’s all about looking like we know what we’re doing. And we know that a 3.5% gas increase is about what the regulators expect. So, go tweak something in the model so we can get to 3.5%.”

My face dropped at the impact of those words. My idealized dreams of how business worked shattered with a single blow. What I thought was a rigorous careful look at data to run the business was anything but.

In the years since, I’ve seen these same factors play out time and again. For example, endless project progress reports meticulously crafted to show progress instead of that same effort being put into actually making progress. Or groups who try to hide when their systems go down so it appears their systems are more stable than the groups who meticulously report their downtime.

One of the unending fights of management is to pierce the illusion to understand what’s really going on. The illusion isn’t all bad. This huge econometric model served a purpose, to allow the company to increase rates more than they would be able to otherwise. They saw real value to them in convincing the regulators they knew their craft well. I would argue that the world would be better served without needing such artifice, but one must also recognize that this is the world we live in.

The point is, jobs were created for specific reasons. Understanding the real reason why work is being done can help you make sure the work is done well. Digging a bit deeper about why a job is really doing what it’s doing can help the company make sure these jobs are done well.