“The truth, the whole truth, and nothing but the truth.”
This famous statement, originating in medieval English courts, has three parts for a very good reason. If you are not telling the truth, you are lying by definition. If you add imaginary details on top of the truth, you equally present a false picture. But when you testify in court, you have to do more than just tell the truth and nothing but the truth: You have to tell the whole truth.
One can create a false impression even when 100% of their words are accurate. When someone selectively omits some essential details, the picture they portray can be radically different. Identifying false, unreliable, or purely invented data is hard enough. When we face accurate but partial data, the danger of falling victim to fake content becomes even more prominent.
Illustration by Tom Wyssocky
The average annual salary in the US rose by 7.6% in 2022. If I want to claim that US employees are in great financial shape and have nothing to complain about, I can use this data to support my point. As a reader, you might think this raise is too high or too low, but no one can argue that employees are not better off on average now than in 2021.
But, of course, this piece of information, as accurate as it is, is only part of the picture. When you drill into this number, you realize that a significant contribution to the average raise originates in people who changed jobs. For people who stayed in their jobs, the average raise was only 5.8%. It is still a raise, but significantly lower than the overall average. That’s not all, though. Inflation rates dramatically affect what you can do with the money you earn, and as it happens, the US inflation rate in 2022 was 6.5%. In absolute terms, employees who didn’t switch jobs made less money by the end of 2022 than in 2021. The claim that employees are better off today than a year ago has become questionable at best1.
Adding two additional bits of information has radically changed the conclusion or at least highlighted an important aspect I initially chose to ignore to prove my point. We can certainly continue to search for other relevant parameters either to understand better the data (like the effect of gender, industry, and geography on the wage increase) or highlight new aspects that are not part of this dataset. And that makes “the whole truth” a tricky standard to meet and Cherry Picking a challenging fake-creation method to deal with.
No one can (or should) mention every bit of data available in the context of every discussion. Doing so is both impossible and impractical. When we say “the whole truth,” we aim for the data and information relevant to the topic we discuss. Whether some data is relevant is a subjective decision: What we consider relevant can be argued by others to be utterly unrelated to the discussion and vice versa. In that sense, any story, any piece of data, and anything we say is, by definition, subjected to Cherry Picking.
So, the question is, are we missing some information that can challenge the argument (and sometimes even invalidate it altogether)? When someone picks only the bits of data that fit their thesis, we are in the realm of fake, whether they do it intentionally in bad faith or are just naive and negligent in their research. In both cases, the impact on the audience can be equally harmful. You don’t have to lie to mislead; You just have to pick the subset of data that supports your case.
Cherry Picking is super easy to take advantage of, and, as we will see in future parts of this series, many other fake-generation methods are built on top of it. When you selectively use the data supporting your thesis, you practically omit any nuance and context. Your thesis becomes shallow, and you fail to provide depth. With the rise of social media, where content is brief by definition, and the attention span of the audience is extremely short, Cherry Picking has become one of the easiest ways to spread fake: You don’t have to invent anything; you can settle with a brief quote of 100% accurate data while omitting what doesn’t serve your needs.
If you genuinely care about the subject and your audience, avoid Cherry Picking. As an audience, we should take special care to protect ourselves from falling prey to this technique.
How to Deal with Cherry Picking
When dealing with Cherry Picking, the amazing accessibility of information on the Internet is a blessing. In most cases, anyone who cares about finding relevant bits of information can do so with nothing more than minimal search capabilities.
When we read a piece of data like the one I used above, the first thing to do is to look for its source (or an equivalent source providing similar information). In many cases, the source of information is more extensive than the quoted bit and can therefore add additional relevant data and nuances.
Other sources of information, sometimes even from adjacent domains, might be relevant to the topic. In the example above, the inflation rate wasn’t mentioned in the first resource, and I had to look it up on other websites. So, the next step would be to consider what else might affect the thesis and look it up. This is far from trivial, especially when dealing with less familiar topics. But once again, the abundance of available information can point us to the relevant aspects we should consider.
For each additional piece of information we find, we should ask ourselves whether it affects the conclusion, add nuance, or raise further questions. Of course, it is up to us to pursue these opportunities, find out more, and eventually reach our own conclusion. Alternatively, we can look for different opinions that already consider these pieces of information.
Yes, identifying and dealing with Cherry Picking is anything but effortless. As I said, the people who generate fake content have the upper hand. But the battle is not lost. When mindful of it, we can find additional pieces of the puzzle to form a better, more complete, and informed picture of reality.
- The data above is brought here just to illustrate the mechanics of Cherry Picking. Since this is not an article about US economics, I haven’t validated or cross-checked this data with other resources. ↩︎