What is Truth to This Scientist?
It was the first few weeks of my first job as a scientist. I was an intern at the U.S. Geological Survey in Menlo Park, CA. As a newly-minted geologist, I was eager to “get out into the field” — work out of doors, searching for useful information on rocks and landscapes. But none of my mentors were going to the field for a couple of weeks, which meant my outdoor training was on hold. Instead, I spent my first days weighing rocks in a dusty room that doubled as a lab and equipment storage facility.
Day in, day out, I stood in the weak sunlight filtering through the grubby window weighing dozens of rock samples both dry and then soaked with water so we could use this information to calculate their density. There was a backlog of samples waiting to be measured, so I barely made a dent each day in the pile of boxes filled with rocks.
I would later learn how important this information is to telling good stories about geology, but at the time it was mind-numbing work with little meaning to me. Nevertheless, I followed the protocol my mentors had taught me. Make every weight measurement twice and write down both numbers on my datasheet. Why? I might misread the scale, or transpose two numbers as I write it down, or a little piece of the rock might fall off in the water as I soaked it for the wet weighing. Making the weight measurement twice ensured I would catch these mistakes.
However, I never seemed to be able to produce the exact same number for each measurement twice. They were often only off by only a fraction of a gram, but they were different nonetheless. Doubt started leaking into my mind. Shouldn’t it be easy to repeat such a simple measurement as weight? Some of the samples I weighed many times to check, but I almost always got slightly different numbers. Was I doing this right? Would the differences matter for our interpretation of this data? But since I saw no reason the scale should be malfunctioning, I just wrote them down and hoped they were ok.
Finally, my mentor reviewed the data I had produced. To my surprise, she responded positively to my numbers saying, “Oh good. Looks like your numbers are close but not exactly the same. Our last intern kept producing weights for each sample that were exactly the same each time. I knew he wasn’t weighing every sample twice but just writing down the same number.” Obviously, the last intern had experienced the same doubts I’d had.
At that moment I learned a fundamental lesson about what real, believable data looks like to a natural scientist. Nature is not precise. Trees never grow perfectly straight. Gravity is not the same everywhere on the planet because the Earth is not perfectly round or perfectly uniform in composition. Birds don’t migrate the exact same flight path every season.
So too, measurements of nature are not precise. Real scientific data lacks absolute certainty, but is instead brimming with truth. When we see a 95% similarity between different measurements, surveys, or observed behaviors, then we have something good. Then we can have confidence that the data is predictable by nature’s standards and we can develop a story to explain it. We cannot ever precisely explain every little variation in such data, so our stories are always to some extent incomplete.
From this experience with the natural world, I have come to believe that imperfection and slight unpredictability are more real and believable than absolute certainty in almost any situation. And that applies to us as scientists too, because we are creatures of the world. Nature dictates that, just as each robin looking for a bug to snack on in my garden has a slightly different technique, we scientists will not all act or think in exactly the same way. But if together we can come to 95% agreement about a story, then we’ve got something good. Then we have truth.
Megan Anderson is a geophysicist and freelance science writer.