Computational Thinking Is Under-Appreciated
An essential 21st century skill that must be encouraged.
Background
“Coding / programming is dead!” The headlines are having a field day proclaiming this diatribe. They are claiming that any time now, we would be conversing with AI and instructing it to get work done, dispensing with the need to write programming codes. Computer scientists, and perhaps even data scientists, would enter a sunset era.
And parents are rightfully concerned about whether they should continue to send their kids to programming school to give them a leg up. The short answer is a resounding “yes”. I am a big believer that coding is an essential 21st century skill. Regardless of whichever career you pursue, coding is a foundation for computational thinking, a necessary mental perspective to surviving and thriving in an AI-enriched world.
And so I dedicate my 74th article to unpacking what is and why the need for computational thinking.
(I write a weekly series of articles where I call out bad thinking and bad practices in data analytics / data science which you can find here.)
What is Computational Thinking?
From Wikipedia: “Computational thinking (CT) refers to the thought processes involved in formulating problems so their solutions can be represented as computational steps and algorithms.” CT is more than just mathematical computing, but rather, it’s a way to think about how computers could accomplish certain tasks from iterations and algorithmic sequences. In layman’s terms, it’s about thinking like a computer.
Excel spreadsheet is arguably one of the best ways to get into computational thinking. Of all the not-so-great products that Microsoft has put out over the years, this venerable granddaddy just cannot be unseated. It’s got a stranglehold on how people think about data problem-solving since 1985. The spreadsheet format is intuitive, even if the Microsoft menus aren’t! The spreadsheet is NOT a database. The spreadsheet forces us to think in deconstructive linear ways; we tend to use Excel where each role uniquely represents an entity (e.g. sales item, customer, employee, etc.), and each column represents an attribute of the entity (e.g. price, age, tenure, etc.) Very often, we use the columns as a “staging area” to compute intermediate data using Excel’s collection of formulae. This staging is a form of CT. Every child should be exposed to Excel. No knowing Excel is like not knowing how to connect via email in this current era.
Interestingly, building with Lego is also a form of computational thinking. It strengthens spatial reasoning. Research has shown that kids (regardless of their gender) who play-build with Lego bricks end of up having stronger Maths and CT abilities.
Importance of Computational Thinking
(Data) Sensemaking, Systems Thinking and Computational Thinking make up the bag of meta-cognitive skills that every person should be equipped with to successfully navigate an AI-enriched, and perhaps, an AI-dominated world. CT is essential for problem-solving and solutioning, and not just in the data domain.
Computational Thinking can be further deconstructed into 4 subcomponents: decomposition, pattern recognition, abstraction and algorithmic thinking. The first 3 do not require a hard data orientation. The first 3 are infinitely transferrable to almost any situation.
I recently attended a Tableau (market-leading data visualisation software) customer event in which they introduced their latest cloud-based feature called “Pulse”, where a user can set up monitoring based on a pre-defined set of metrics and have it published automatically. What struck me the whole time was how solution vendors are constantly trying to dump down the need for CT by shielding it from the user. This eventually leads to a disconnect where the users don’t really understand the core design and computational workings of the solution, and therefore, unable to extract its full utility. Instead, we should be expecting more from the user. Every well thought through solution should enhance the user’s skillsets.
But Computation Thinking Will Evolve
I can hear some of you saying, “Why would I require CT if I can simply tell the computer (AI) what I want done?”. Think about how you would instruct a fellow human to get a piece of work done. You have to explain in a language and context of the person doing the work. Even then, there are many times when the person gets it somewhat wrong, and you wonder if it’s the person’s fault for not understanding the instructions, or it’s your fault for not communicating it well. That will be the same principle that applies in our interaction with AI. How do we help the AI improve its accuracy of interpreting what we want if we cannot appreciate how the AI mechanism works?
Consider the case of prompt engineering. This is an extended form of computational thinking. Inputs are now in short (key) words / phrases. Similar words can elicit very different outcomes. Understanding the tokenisation and transformer (attention) mechanism of Gen AI is thus important to get the most accurate outcomes or troubleshoot unintended outcomes.
Conclusion
CT is an evolving, dynamic domain. Computer interfaces and algorithms will continue to evolve. As we learn to “think” like computers, computers are “learning” to think like us. Perhaps the two will converge at some point :)
