A mote of dust floating isolated in a glass of water will move randomly in every direction, buffeted by the natural motion of its surroundings. Over time, however, those random movements will compound into a significant drift in a certain direction. Physicists study this phenomenon and call it “The Drunkard’s Walk,” but that averaged, long-term movement in a particular direction is something much more important to the marketing world: it’s a trend.

In his book The Drunkard’s Walk: How Randomness Rules Our Lives, Leonard Mlodinow introduces and explores this and other theories of randomness and statistics that affect our lives and then applies those theories to everyday situations.

Mlodinow demonstrates and discounts common fallacies when it comes to judging random chance. In advertising, one of the biggest quandaries is the extent to which we should rely on statistics and trends. Though it’s true that a random sampling of the population will sometimes represent the greater whole, Mlodinow references a study of preference in musical tastes that confounds the very idea of how we gather data: focus grouping.

Essentially, a team of researchers set up a microsite in which several groups of test subjects could listen to the same 40 tracks by completely unknown artists, rate them, and then read reviews and post their own. In each of these groups, the reviews and ratings could only be read by members of the same group.

Our intuition tells us that over a large enough sampling, each of the smaller communities would show the same tracks as most popular, or there would at least be a noticeable correlation among the highest rated songs. Instead, they found wildly differing results—that a track rated #1 in one community could be rated #40 in another—and they were lost as to why such an anomaly had formed.

As it turns out, the populations in their groups weren’t even following their own tastes, because songs that started out popular in certain groups tended to continue to be popular based entirely on the tastes and reviews of others. People weren’t judging, they were subconsciously following the trend.

If we had been privy to just one of those groups, however, we would have assumed a pattern and considered the top song to be just that—the song that would be most popular overall. But we would be wrong.

Anyone looking to drive consumers to their brand could consider these unpredictable, unknown variables as a setback. After all, if the future is unforeseeable and chance is inescapable, how can we be sure that we’ll be successful? Well, we can’t. But knowing that—and accepting it—is the first step to making that setback a positive result.

We shouldn’t rely too heavily on our statistics, because we must always consider the unpredictability of human nature when we consider the predictability of success. Mathematicians and physicists may try to predict the path of a particle in a fluid suspension, but a mote of dust is a mote of dust. When dealing with human trends, the future is even more uncertain.