If you've watched Pluribus, Vince Gilligan's Apple TV+ series, you already understand synthetic data. You just don't know it yet.
Here’s the setup. Something spreads across the planet and joins almost every human being into one connected mind. Billions of people, one shared consciousness. People become calm, agreeable, and eager to serve. There are a few people, like Carol, who are somehow immune to this “thing.” Carol suddenly finds herself in a world where everyone is technically still there but nobody is their own person anymore.
People essentially become a connected system with access to everyone’s information, which is the best illustration of synthetic data I have come across.
What is Human Data? Or What Was the World Like Before the Virus?
Before the virus took over the earth in the show, everyone had their individual knowledge, experiences, and opinions. People disagreed on things, they argued.
Think of a bell curve. Most people cluster near the middle, and the unusual cases thin out toward the edges. Carol, for example, is an independent thinker whose opinions sit closer to the edge than anyone else’s.
What is Synthetic Data? Or What Was the World Like After the Virus?
When the virus spread, it didn’t take away the knowledge. The hive mind still remembers everything humanity knew. However, it takes away independence and uniqueness, with many opinions merging somewhere in the middle, into one.
In practice, synthetic data consists of records a computer generates to stand in for real ones. It creates personas that are synthetic from what it learns about opinions of real human beings that are put into different categories, patients, businesspeople, etc.
Synthetic data, simply put, is an average of many opinions that existed or were recorded in the past, smoothed out. The data generated in this way reproduces the patterns it learned, which means it drops the oddball cases that showed up only once or twice. With every pass it grows more confident about a steadily narrower world.
Like Zosia, the character from the show affected by the virus, the data has access to everything the world collectively knows. And like with Zosia, whatever was specific to any one person got smoothed away.
Think About a Survey
Synthetic data is a real tool. It’s not a shortcut, or a fraud, as skeptics sometimes label it. It is a great answer when you need volume, or when you are short on time, or budget. Use it, and you are tapping the same collective mind the virus created in Pluribus.
The thing to consider when choosing it, however, is where on the curve would the potential answer you are looking for live? If the answer you are looking for lives near the middle of the curve, synthetic data may be exactly what you need. If the answer you are looking for, however, depends on these edges, you may need to seek out a Carol to get the information you need.
Putting it in show’s terms: the hive is remarkably good at answering questions. What it is still struggling with, however, is noticing something new that hasn’t been said before.
If you haven’t, watch the show. I tried not to give away too many things for a reason!