Have you heard of sampling? To not get too technical about it, it’s a way to understand how many people prefer a particular thing without asking every person in existence. The idea is, mathematically you can determine how many people you need to sample to achieve a high likelihood of being able to extrapolate their answers across the population you want to know the answer to.

For instance, I could ask “what percent of the population prefers Vanilla over Chocolate ice cream?” And I could sample around 2000 people and get a definitive answer to that question for the USA (statistics folks will note it may not even take that many, but I am a student of statistics, not a practitioner, I leave that to far more capable people than myself).

You may think there’s no way 2000 people are representative of a population of 330,000,000, but depending on how the sample is chosen and depending on the questions and depending on the answer rate, it’s not only possible, but you can answer it with a 2.6% error rate, which is great considering you’d go bankrupt long before polling 330,000,000 people.

But that’s not why I bring it up. I bring it up because if those numbers shock you, and you think that clearly it’d be more accurate to always poll more people, you’d be wrong.

More than that, though, the part that’s interesting is that we do that all the time in software. We take our expertise and try to apply it to things we aren’t experts about. You see it in UX discussions, you see it in design reviews, you see it just about all facets of our work. If you’ve ever claimed SQL is superior to NoSQL or vice versa — without defining the context you’re talking about, you’ve even committed this fallacy.

When we step back and realize we aren’t experts even though we are smart, we allow room for the experts to help guide us. That silence is worth it.