Dark side of the (data) force #1: Fear

At Spike we are used to frequently read internet articles and academic papers showing all the great stuff that data scientists can do using machine learning and other advanced analytics techniques. We even have a weekly session to admire and learn from others’ work called La Escuelita — The Little School 🤓 — in which someone suggests a paper, article, blog post, or lecture that we then study and discuss in detail as a team. However, less is shared about the not-so-great parts of our work as data scientists, machine learning engineers, and, more in general, data practitioners.

Just like in most social networks: when people share and celebrate only the great stuff (hiding and filtering the mistakes and hassles of real life), it’s very likely that some will feel excluded and guilty of not living up to others’ awesomeness which, in turn, can elicit a dark side full of anxiety, fear, false pride and lack of self assurance.

That’s why in this series of articles I want to talk about…

The dark side of the (data) force!

🎻Duh Duh da da da Duh da da da Duuh🎶

So, let’s kick off with the one that I think is most widespread: fear 😰.

Fear: the path to the dark side

Have you ever felt that other members of the data community of practitioners are smarter and more talented than you? Or that other companies are doing really cool stuff in comparison to what you do (which is too simple and easy)? That you don’t know enough statistics? That you are not good enough writing code? Or maybe that someone will, at some point, find that you don’t belong to where you are because you are not prepared enough?

Well, if you have felt any of these symptoms then it’s very likely that you have suffered from the impostor syndrome.

I won’t go over the details of it, but a short (and very personal) summary would be: “the feeling that almost every smart and skilled person feels in their guts of not being ready nor qualified for the task or not belonging to a certain group of talented people”; in other words, feeling like an impostor that will be shamely uncovered sooner or later.

Impostor Syndrome is a feeling that almost every smart and skilled person feels in their guts of not being ready for the task or not belonging to a certain group of talented people

At Spike we have confirmed this over and over again: almost everybody thinks that the rest of the team knows more, has more talent, and is more prepared for advance analytics work. Part of my daily job at Spike is to remember to some of the best professionals I have ever worked with, how awesome they are and confirming that they really belong to the team. But here is the paradox: I have the exact same feeling every day! 😳 (Don’t tell anyone, but after years of practice I still have trouble distinguishing type I from type II errors, normalization from de-normalization, writing basic regular expressions… and git is still too awefully complicated for me to use it from the command line).

This widespread fear of being an impostor has terrible consequences:

• Lack of security. To participate, to give new ideas, to ask questions, to say that you didn’t understand something and asking for help. In sum, you become a reduced version of yourself. Your talent and creativity gets multiplied by a number between 0 and 1.
• Loosing your own style. As you want to belong to the collective of awesome people, it’s easy to assume that your own style of doing your work is wrong and embracing others’ style is just right. In sum, losing a great part of what makes us unique.
• Difficulty to assume vulnerability. As everyone feels fear of being an impostor, nobody wants to assume and show any weakness which, in turn, exacerbates bragging about awesome things and sidestepping not-so-cool ones. In sum, hiding what makes us humans and what help us the most in connecting with others.
• Inaction from not feeling ready yet. I’ve seen a lot of talented professionals that declare that need more understanding and preparation — reading a new book, taking another online course, getting another certification — before starting doing actual stuff; they don’t feel ready just yet. Well, in my experience, there is no better way of understanding something (a problem, a discipline, an algorithm, a programming language) than by rather… doing. Then, inaction blocks learning which, in turn, blocks action.

Once you become a reduced version of yourself (or your team a reduced version of itself), then is very easy to get deeper into the dark side in a reinforcing loop: poor results induce more fear which, in turn, induces even less creativity, uniqueness, action, learning, and so forth .

In sum, fear might hide your talents, reduce your creativity and uniqueness, destroy social connection with others, and block action and learning. Once you become a reduced version of yourself (or your team a reduced version of itself), then is very easy to get deeper into the dark side in a reinforcing loop: poor results induce more fear which, in turn, induces even less creativity, uniqueness, action, learning, and so forth .

Why the data community is full of fear?

First, because working with data mixes both technical with creative skills in a very intensive way. The technical side of working with data and machine learning is more obvious, but to understand its creative side consider two typical questions in advanced analytics: which new features could you build for your predictive model? What should we do with the model once it’s trained in order to get to, say, reduce churn? If you think about it, those are eminently creative questions. Technique helps creating a map (i.e., a non-perfect model of a territory that is being explored), creativity is the one that dictates where to go within that territory.

Technique helps creating a map (i.e., a non-perfect model of a territory that is being explored), creativity is the one that dictates where to go within that territory.

On the technical side, fear is related to the amount of information available that has not been internalized by someone. For instance, most programmers don’t often declare publicly something that we all do: googling almost everything that needs to be done (thank you, Stack overflow. Thank you, Quora). So, for someone who is getting started in coding, it’s very easy to assume that you need to know the methods and syntaxis of every programming language before being productive with one. Something similar happens with statistics and probability, ML algorithms, among other fundamental building blocks of data science. Inaction from not feeling ready.

On the creative side, well… insecurity and fear is perhaps the one and foremost common feeling among creators and artists. “This code looks like crap, people will hate it”, “my song sounds awful”, these are things that even the most important programmers and songwriters feel during their creative process. To create something new is, ultimately, to overcome those feelings of shame and fear, and to make something unique, singular, and far from perfect.

As Javier Cercas once said: “courage doesn’t consist in not being afraid — that is recklessness — but rather in mastering it, doing what you have to do and move on”

What can we do about fear?

As Javier Cercas once said: “courage doesn’t consist in not being afraid — that is recklessness — but rather in mastering it, doing what you have to do and move on”. Artists and creators are masters of dealing with fear. In the same way, data scientists and machine learning engineers should be masters in starting doing stuff without understanding everything that is involved; masters of asking for help; masters of showing vulnerability; masters of cultivating their own, unique style; ultimately, masters of curiosity and humility.

So, next time you find yourself typing “how to…” in google, bare in mind that the force, and us, will be with you.

Thanks to Rafael Sacaan, Diego Aguayo, Emilia Fresard, Martín Villanueva, and Juan Pablo García.

Spike is a data innovation lab that empowers large organizations with the adoption of AI and data-driven value exploration, building tailor-made solutions to complex problems using large data volumes and machine learning techniques, across multiple industries like retail, airlines, health, and mining.