Your mind is only as good as its training data

Your mind is only as good as its training data
Photo by Pietro Jeng / Unsplash

This series explores proven Machine Learning principles that we can apply to our own mindset for a better life. Part 1 is below. I’ll update this article with links to future installments.

Machine Learning is the study of computer algorithms that improve automatically through experience.

Do you know what else is supposed to improve automatically through experience? Your mind.

People fawn over celebrities and business leaders alike, aspiring to copy their playbooks for success. Why not learn from the playbook that is being used for cancer detection, fraud detection, computer vision, sentiment analysis, self-driving cars, or behavioral analytics? If any human possessed the specific knowledge of a well-trained model, he or she would probably win the Nobel Prize in their field.

Let’s learn together from the principles that optimize the world’s most powerful algorithms and models.

Part 1) Sampling bias / Bad training data

— be intentional about what you expose yourself to
— understand your blind spots
— take broader samples

When we are watching the news we assume that their mission is to give us a concise presentation of the most important events going on. News outlets serve their advertisers, not their base. After all, the advertisers want eyeballs, and the news outlets have figured out that it's more lucrative to keep you afraid and glued to the television for faux-crisis after faux-crisis. Not only are we less informed, we feel the need to consume more and more mindless information that doesn’t have any effect on our daily lives or our primary mission.

The news is less like
“Here’s what you need to know so you can be on your way.”

And more like:
“Look at all the things going wrong.”
“Look at those weird people over there.”
“Stay tuned. Always.”

Every time you consume a piece of information, it imprints in your brain whether you realize it or not. That means that the information filters we choose have a profound effect on the way we think and the way we behave.

There is plenty of footage of cops beating peaceful protestors.
There is also plenty of footage of people rioting, looting, and assaulting unprovoked.

If one group of people sees 98% cop beatings and the other group views 98% riots and looting, it’s safe to say that those groups will have different world views. If these groups don’t realize how they are being manipulated against each other, it sets the stage for a dangerous showdown.

Another example, being a person of color who wants to enjoy the outdoors, I have been conditioned by the media to be scared of the outdoors. It’s only news when a minority hiker is attacked outdoors; we don’t ever hear about the thousands of hikes that happen without incident. What’s the actual, realistic likelihood of me getting attacked outdoors while hiking or camping?

Understanding your blind spots

@icons8 on Unsplash

There is only so much time in the world and there are so many information filters to choose from. Each person will undoubtedly have blind spots where information is lacking. The easiest example of this is the political media landscape. If you only subscribe to Fox News you will hear a flurry of Right-Good, Left-Bad commentary. Inversely, MSNBC viewers get bombarded with Left-Good, Right-Bad commentary. Each cohort has two blind spots; Fox News viewers don’t hear about anything Left-Good or Right-Bad and MSNBC viewers rarely have to contemplate Right-Good or Left-Bad.

The harmful result of this is that people are shouting past each other. Let’s take the polarizing issue of regulation vs deregulation.

Right-Good, Left-Bad: Academics and lawyers who know nothing about our industry will create restrictive regulations with plenty of loopholes. These same people will leave the government and earn millions in consulting fees for helping a privileged few companies exploit those very same loopholes.

Left-Good, Right-Bad: Big Bad Business wants to run rampant with our land and they don’t consider the damage they are doing to the taxpayer. We will create regulations that ensure fair use in accordance with the best interests of the taxpayer. Whoever opposes these regulations doesn’t care about people, they only care about exploiting for profit.

Notice that anti-corruption is a bipartisan issue, but each camp has a different cohort in its crosshairs. This is only possible if we don’t consider all viewpoints.

Should people drafting regulations be allowed to help companies exploit those same regulations?
When should we use regulations vs allowing the court system to run its course?

None of the important issues get debated since we don’t see the entire picture.

Take Broader Samples

Expose yourself to multiple data points and angles. Every outlet has a bias. There’s simply too much information out there. Decisions are made about what to share and what to leave alone, even if the decision-maker attempts to be neutral about the content. A more productive endeavor would be to discern each outlets’ mission or raison d’être. Fair and Balanced doesn’t pay the bills anymore. One mitigation strategy is to subscribe to individual reporter’s newsletters or Twitter accounts. That way you begin to develop your own point of view instead of accepting the mainstream framing of each and every situation.

Stay Tuned

Future installments will cover overfitting, model selection, outliers, and exploration vs. exploitation.