Why Everyone Should Learn to Code — The Serious CTO Deep Dive (FFFC 34)
---
SECTION 1: THE CLAIM AND WHY IT MATTERS
"Everyone should learn to code." It is one of the most repeated, most contested, most misunderstood statements in modern tech culture. Former US President Barack Obama said it. Microsoft founder Bill Gates said it. Bloomberg's founder Michael Bloomberg famously attempted to learn it, failed, and then funded initiatives promoting it anyway. The slogan became a bumper sticker, a hashtag, a curriculum mandate in schools from Estonia to Australia, and — for a significant portion of the software development community — a source of eye-rolling frustration.
But here is the thing: the people rolling their eyes are usually misunderstanding the claim. And the people making the claim are usually doing a terrible job of explaining what they actually mean. This document will pull apart the argument from first principles, examine the evidence honestly, and help software developers understand why this debate matters deeply to their careers, their organizations, and the future of the discipline they have chosen.
The "learn to code" argument is not really about whether everyone should become a professional software engineer. It never was. The deeper argument is about computational thinking, digital literacy, automation awareness, and the long-term economics of labor in a world where software is eating everything. That is the argument worth taking seriously.
---
SECTION 2: DEFINING THE TERMS — WHAT DOES "LEARN TO CODE" ACTUALLY MEAN?
Before we can evaluate the claim, we need to nail down what it means. There are at least four distinct interpretations in circulation, and conflating them is the source of most of the confusion in this debate.
**Interpretation 1: Vocational coding — become a professional developer.**
This is the interpretation most professional developers assume when they hear the slogan, and it is the one they are most likely to dismiss. Not everyone should become a software engineer, and the argument that they should is indeed weak. Coding as a profession requires not just syntax knowledge but systems thinking, debugging discipline, architecture awareness, toolchain management, and years of accumulated pattern recognition. You cannot shortcut your way into senior engineering competency with a 12-week bootcamp any more than you can shortcut your way into performing surgery. This interpretation, taken literally, is wrong.
**Interpretation 2: Computational thinking — understanding how software systems work.**
This is the interpretation that the most credible proponents actually mean. Jeannette Wing, the computer scientist who popularized the term "computational thinking" in her landmark 2006 Communications of the ACM paper, defined it as "the thought processes involved in formulating problems and their solutions so that the solutions are represented in a form that can be effectively carried out by an information-processing agent." This is not about writing code. It is about decomposing problems, recognizing patterns, designing algorithms, and thinking about abstraction. This is genuinely useful for virtually every knowledge worker.
**Interpretation 3: Scripting and automation literacy — the ability to write small programs to solve personal or professional problems.**
This is the pragmatic middle ground. Can a marketing analyst write a Python script to clean a dataset? Can an accountant write a macro to automate a repetitive spreadsheet task? Can a journalist write a simple web scraper to gather public data? This level of coding literacy is achievable by most people with moderate effort, and it delivers concrete, measurable returns on time investment.
**Interpretation 4: Digital literacy — understanding technology well enough to participate meaningfully in a digital society.**
At this level, "coding" is almost a metaphor. The argument is that citizens who do not understand how algorithms work, how data is collected, how software systems make decisions, and how digital infrastructure operates are fundamentally disadvantaged — economically, politically, and socially. This is less about syntax and more about epistemic independence.
The most defensible version of the "everyone should learn to code" argument draws on interpretations 2, 3, and 4. Interpretation 1 is a straw man, but it is the one that keeps getting attacked.
---
SECTION 3: THE ECONOMIC AND LABOR MARKET REALITY
Let us look at the numbers, because this is where the argument gets its real force.
The US Bureau of Labor Statistics (BLS) Occupational Outlook Handbook reports that employment in software development and quality assurance is projected to grow 25% from 2022 to 2032, much faster than the average for all occupations. The median annual wage for software developers was $127,260 in May 2023. There are approximately 1.8 million software developers currently employed in the United States alone.
But the more interesting data set is not about software developers. It is about what happens to non-developers who gain coding skills.
A 2016 study by Burning Glass Technologies (now Lightcast) analyzed 26 million US job postings and found that jobs requiring coding skills pay an average of $22,000 more per year than jobs that do not in the same field. A marketing manager who can code earns significantly more than one who cannot. A business analyst who can write SQL queries is dramatically more productive and better compensated than one who cannot. The same pattern holds in finance, healthcare administration, journalism, law, and operations.
According to the World Economic Forum's "Future of Jobs Report 2023," technology literacy, AI and big data skills, and programming are among the top skills expected to see increasing demand through 2027. The report surveyed over 800 companies across 27 industries and 45 economies. It found that 65% of employers expect AI and big data proficiency to increase in importance, and that programming and technology literacy are in the top ten most sought-after skills for the coming decade.
McKinsey Global Institute's "Skill Shift: Automation and the Future of the Workforce" (2018, with updated analyses through 2023) estimates that by 2030, as many as 375 million workers globally — about 14% of the global workforce — may need to switch occupational categories as automation and AI reshape labor markets. The workers best positioned to navigate these transitions are those with what the report calls "technological skills" — including basic coding and data manipulation abilities.
The Stack Overflow Developer Survey 2023 — the largest survey of developers in the world, with over 90,000 respondents — found that the median salary for a developer in the United States is $150,000. But more relevantly for the "everyone should code" argument, the survey also found that a significant portion of developers (roughly 17%) identified as "part-time" or "hobby" coders whose primary profession was something else entirely — product management, data analysis, design, operations, and research.
GitHub's "State of the Octoverse 2023" report noted that the platform surpassed 100 million developers globally for the first time. It also reported that the fastest-growing segments were not traditional software engineers but domain experts learning to code — scientists, educators, journalists, and policymakers who were building tools for their specific fields.
---
SECTION 4: THE COMPUTATIONAL THINKING EVIDENCE BASE
Jeannette Wing's 2006 paper in Communications of the ACM is the foundational text here, but the evidence base has grown substantially since then. Her core argument was that computational thinking should sit alongside reading, writing, and arithmetic as a fundamental analytical skill for the 21st century.
A 2019 meta-analysis published in the journal Computers & Education examined 105 studies on coding education for non-programmers and found statistically significant improvements in problem-solving ability, logical reasoning, and academic performance in mathematics after coding instruction. The effect sizes were moderate but consistent across age groups, subjects, and instructional contexts.
The UK government made computing a statutory subject in its national curriculum starting in September 2014, making it one of the first countries to mandate coding education from primary school level (age 5) through secondary school (age 16). The Royal Society's 2017 report "After the Reboot: Computing Education in UK Schools" found early evidence that students who engaged with programming concepts showed improved abstract reasoning and were better at breaking complex problems into manageable components.
Estonia has been teaching programming in schools since 2012 through its ProgeTiiger initiative, and by 2021, the country ranked in the top tier of European nations for digital skills. Estonia's e-government infrastructure — which handles everything from voting to tax filing to healthcare records — was built on a culture of digital literacy that started with exactly this kind of early-age coding education.
Finland integrated computational thinking into its national core curriculum in 2016 as a cross-subject competency rather than a standalone computer science subject. Finnish students learn to think algorithmically in the context of mathematics, science, arts, and even physical education. Early assessments suggest this embedded approach yields stronger transfer effects than isolated coding instruction.
---
SECTION 5: THE CASE FOR DEVELOPERS — WHY THIS DEBATE IS NOT JUST ACADEMIC
Here is where professional software developers should pay close attention, because the "everyone should learn to code" movement has direct implications for how you work, who you work with, and what the competitive landscape of software development looks like.
**Implication 1: The rise of the informed non-technical stakeholder.**
If product managers, executives, legal teams, and operations staff have basic coding literacy, they become dramatically better collaborators for engineers. They understand why certain features are hard. They can read a data schema. They can evaluate a technical trade-off without being exclusively dependent on the engineering team's framing. This is almost universally considered a good thing by experienced developers. One of the most common complaints in developer communities — on Reddit, Hacker News, Stack Overflow, and in conference hallways — is that non-technical stakeholders make poor decisions about technical systems because they fundamentally do not understand what they are asking for. Broader coding literacy directly addresses this problem.
**Implication 2: The democratization of simple automation raises the floor.**
When business analysts can write Python, and when marketing teams can use SQL, and when operations teams can wire together API calls with tools like Zapier or Make (formerly Integromat), the demand on engineering teams changes. Engineers get freed from low-value automation work and pushed toward genuinely complex problems. This is a productivity multiplier at the organizational level. A 2022 study by the MIT Sloan Management Review found that companies with "democratized data and analytics capabilities" — including basic coding access for non-engineers — made decisions 2.4x faster and reported significantly higher employee satisfaction among both technical and non-technical staff.
**Implication 3: Low-code/no-code is not a threat — it is a forcing function.**
The rise of platforms like Airtable, Retool, Bubble, and Webflow has genuinely enabled non-developers to build functional applications. Gartner predicted in 2021 that by 2024, low-code application development will be responsible for more than 65% of application development activity. That prediction has largely held up. But what this has actually done in practice is not replace developers — it has redefined the value of developers. When non-developers can handle the routine, the developers who remain in demand are those who can architect the systems that non-developers build on top of, solve the edge cases that no-code platforms cannot handle, and evaluate the security and scalability implications of democratized development. Basic coding literacy among the general population amplifies, not diminishes, the value of deep expertise.
**Implication 4: AI coding assistants are creating a new category of "vibe coders."**
The term "vibe coding" — coined semi-humorously by Andrej Karpathy in a February 2025 tweet — describes the practice of using AI assistants like GitHub Copilot, Cursor, and Claude to write code through natural language prompts, without necessarily understanding the underlying mechanics deeply. This is perhaps the most important current manifestation of the "everyone should learn to code" debate. AI has dramatically lowered the barrier to generating working code. The question is no longer whether you can write a for loop — it is whether you understand what you are generating well enough to know when it is wrong. This makes computational thinking more important, not less. Someone who "vibe codes" without any understanding of what the code is doing is building on sand. Someone who vibe codes with a solid mental model of the systems involved can move extraordinarily fast. The "learn to code" argument, in this light, becomes the argument for developing the mental model — even if you rely on AI for the syntax.
---
SECTION 6: REAL-WORLD CASE STUDIES
**Case Study 1: The New York Times Data Journalism Team**
The NYT's interactive news team has for years employed journalists who code — reporters who write Python and JavaScript alongside traditional journalism. Matt Ericson, Archie Tse, and Amanda Cox built the foundation for what is now considered one of the leading data journalism operations in the world. When the pandemic hit in 2020, the NYT's coding journalists were able to scrape, analyze, and visualize COVID case data at a speed and granularity that non-coding newsrooms could not match. Their interactive COVID tracker became one of the most accessed pages on the internet for two years. The team's success is a direct vindication of the "journalists should learn to code" argument that was heavily debated in the early 2010s.
**Case Study 2: Stripe and the Informed Business Team**
Stripe, widely regarded as one of the best-run engineering organizations in the world, has made it explicit in its culture that product managers, designers, and business operations staff should have strong technical literacy. Stripe's documentation — famous for being the best in the API industry — was written partly by non-engineers with coding literacy who could read and test the APIs they were documenting. The result is documentation so good that developer onboarding time is measured in minutes rather than hours or days. Technical literacy among non-engineers was a strategic differentiator.
**Case Study 3: Kaggle Competitions and the Democratization of Machine Learning**
Kaggle, the data science competition platform owned by Google, has over 10 million registered users. A significant proportion of top competitors on Kaggle are not professional software engineers — they are statisticians, biologists, physicists, economists, and domain experts who learned Python and R well enough to apply machine learning techniques to their domain-specific problems. The winners of multiple Kaggle competitions have been domain experts who understood the problem better than professional engineers, and who combined that domain knowledge with just enough coding skill to implement powerful solutions. This is perhaps the clearest empirical evidence that "enough coding" — not expert-level engineering — delivers disproportionate value when combined with deep domain expertise.
**Case Study 4: Spreadsheet Power Users in Finance**
The financial industry has long been a microcosm of this debate. Excel power users — analysts who know VBA, who understand data models, who can build complex financial models — consistently outperform peers who rely on off-the-shelf tools. A 2020 survey by the CFA Institute found that 64% of investment management professionals cited "Python/data analysis skills" as a top skill gap in their organizations, and that analysts with these skills were commanding 15–20% salary premiums over otherwise equivalent peers. Goldman Sachs made headlines in 2018 when it was revealed that the firm had introduced Python programming as a requirement for its analysts — not because it wanted to turn them into engineers, but because data fluency had become a baseline requirement for financial analysis.
**Case Study 5: The NHS and Clinical Informatics**
The UK National Health Service has been through a long and painful journey with digital transformation. A significant part of its early struggles with electronic health records (most infamously the cancelled £12.7 billion National Programme for IT in 2011) came from a complete disconnect between the clinical staff who understood healthcare processes and the technical teams who built the systems. When NHS began embedding "clinical informaticists" — doctors and nurses who learned enough coding and data modeling to bridge the gap — outcomes improved significantly. NHS Digital's later, more successful implementations relied heavily on clinical staff who could speak both languages. The estimated cost savings from more technically literate clinical teams guiding better technology procurement and implementation run into billions of pounds.
---
SECTION 7: THE CONTRARIAN VIEW — WHEN "LEARN TO CODE" IS BAD ADVICE
No serious treatment of this topic can avoid the genuine criticisms of the learn-to-code movement, because several of them are legitimate.
**Contrarian Point 1: Time opportunity cost is real.**
Learning to code takes time. For a 45-year-old teacher, a 50-year-old nurse, or a 55-year-old small business owner, the time required to reach even basic coding competency may genuinely not be the best investment of their limited learning time. If your profession is not going to be materially improved by coding skills, the argument weakens considerably. The "everyone" in "everyone should learn to code" is genuinely too broad.
**Contrarian Point 2: The skills decay problem.**
Coding is a perishable skill. If you learn Python basics in a weekend workshop and then do not use Python for six months, most of what you learned has evaporated. The maintenance cost of coding skills — staying current with evolving ecosystems, debugging problems that require understanding modern toolchains — is non-trivial. Critics argue that the learn-to-code movement significantly underestimates this ongoing cost.
**Contrarian Point 3: Bad code is worse than no code.**
This is a genuinely important point that professional developers feel viscerally. A non-developer who writes a little bit of SQL, Python, or JavaScript and deploys it in a production context without understanding security, error handling, or performance implications can create serious problems. SQL injection vulnerabilities, exposed API keys, unhandled exceptions that corrupt data — these are not hypothetical risks. They are constant realities in organizations where coding literacy is celebrated without accompanying engineering discipline. The "everyone should code" movement is sometimes accused of producing dangerous half-knowledge.
**Contrarian Point 4: The solution may be better tools, not more coders.**
Technology writer Evgeny Morozov and others have argued that "learn to code" is a supply-side answer to what is actually a demand-side problem. If technology is poorly designed, the answer might be to demand better-designed technology rather than training the population to work around its poor design. Why should every accountant learn to write SQL? Perhaps the answer is better accounting software that does not require SQL. This is a critique worth taking seriously, though it somewhat begs the question — in a world where software is genuinely complex and where general-purpose tools will always have limitations, some level of technical literacy will always be valuable for power users.
**Contrarian Point 5: The learn-to-code pipeline has its own equity problems.**
The learn-to-code movement has largely benefited people who are already relatively advantaged — those with stable internet access, time to invest in self-directed learning, social networks that can recommend resources, and financial cushion to take educational risks. Research by the Kapor Center and others has documented that coding bootcamps, despite their promise of democratization, have frequently failed underrepresented groups at significantly higher rates than their marketing suggests. The "anyone can learn to code" message can, when deployed uncritically, obscure structural inequities that determine who actually succeeds.
---
SECTION 8: COMMON TRAPS AND MISTAKES DEVELOPERS MAKE IN THIS DEBATE
Professional developers are not innocent parties in the dysfunction of this debate. Several specific patterns come up repeatedly in how developers engage with the "learn to code" question.
**Trap 1: Gatekeeping dressed as standards.**
The dismissal of the learn-to-code movement by professional developers sometimes reflects a genuine concern about quality — but sometimes it reflects professional protectionism. When developers argue that "real coding" requires years of study and that anything less is not real coding, they are drawing a circle that conveniently includes themselves and excludes everyone else. This is intellectually dishonest. There is a spectrum from "has never written a line of code" to "senior staff engineer at a top-tier tech company," and claiming that only one end of that spectrum has legitimate standing is gatekeeping, not professional standards.
**Trap 2: Misattributing the argument.**
Developers often argue against the weakest version of the "learn to code" claim — the idea that everyone should become a professional developer — while ignoring the stronger versions about computational thinking and automation literacy. This is a straw man fallacy. Before dismissing the argument, nail down which version is being made.
**Trap 3: Assuming domain expertise and coding are mutually exclusive.**
There is a persistent assumption in some developer communities that deep domain knowledge (in healthcare, law, finance, education, etc.) and coding skill are separate and that one type of person gets one and not the other. The Kaggle example above, along with numerous real-world cases, should disabuse anyone of this notion. The combination of domain expertise and coding competency is extraordinarily powerful.
**Trap 4: Underestimating the value of "enough."**
Developers who measure coding ability on the scale of professional software engineering consistently underestimate the value of "enough to be useful." The difference between a business analyst who can write a basic SQL query and one who cannot is enormous in practice, even though neither is close to being a database engineer. Developers who only measure value at the expert level miss most of the value being discussed.
**Trap 5: Not understanding the collaboration dividend.**
Some developers resist the idea of non-developers learning to code because they fear it will make their jobs redundant. The evidence strongly suggests the opposite — technically literate non-developers make developer teams more productive by reducing translation friction, improving requirement quality, and handling automation tasks that should never have needed a developer in the first place.
---
SECTION 9: THE AI INFLECTION POINT — HOW THE CALCULUS HAS CHANGED
The "learn to code" debate has entered a genuinely new phase in 2024 and 2025, and anyone who does not engage with AI's role in this is missing the most important current dimension.
GitHub Copilot, launched in 2021 and now used by over 1.3 million developers (as of 2023 reporting), has demonstrably changed what it means to code. A 2023 GitHub study found that developers using Copilot completed tasks 55.8% faster than those who did not. More interestingly for the "learn to code" argument, separate research from Microsoft published in 2023 found that non-developers using Copilot-style tools were able to complete simple coding tasks at rates that would have been unimaginable without AI assistance.
Claude, ChatGPT, and similar large language models have further democratized code generation. In 2024, the "natural language to working code" pipeline has become reliable enough for simple to moderate complexity tasks that the barrier to entry has dropped enormously.
What does this mean for the "learn to code" debate? Several things:
First, it strengthens the case for broad coding literacy because it lowers the cost of entry. You no longer need to memorize syntax to get value from programming. You need to understand what you are asking for, evaluate what you receive, and know enough to debug when it fails. That is a computational thinking argument, not a vocational coding argument.
Second, it increases the premium on understanding fundamentals. As AI-generated code becomes ubiquitous, the ability to critically evaluate that code becomes more valuable. Someone who learned computational thinking and has a mental model of how software works will be far more effective with AI coding tools than someone who has no such foundation. The "everyone should learn to code" argument, in the AI era, becomes the argument for developing the minimum viable technical intuition needed to work with AI effectively.
Third, it creates new failure modes. The phenomenon of non-developers using AI to write code they do not understand and deploying it without testing or review is already creating security and reliability problems in organizations. This is the "bad code is worse than no code" argument on steroids. It argues not for rejecting coding literacy but for ensuring that coding literacy includes enough understanding to know the limits of AI-generated code.
Andrej Karpathy's framing of "vibe coding" is useful here. The vibe coder who has enough computational thinking to sanity-check AI output is dramatically more powerful than one who does not. The learn-to-code movement, properly understood, is about building that sanity-checking capacity — not about grinding through textbooks to memorize quicksort implementations.
---
SECTION 10: WHAT DOES THIS LOOK LIKE IN PRACTICE?
This is the most practically useful section for developers thinking about how to engage with this topic — whether in hiring decisions, mentoring non-developer colleagues, contributing to curriculum discussions, or simply forming an opinion.
**For individual non-developers (what to actually learn):**
The practical recommendation that holds up across most serious analysis is a tiered approach. Start with data literacy — understanding spreadsheets deeply, including functions, pivot tables, and basic statistical concepts. Move to SQL — the ability to query a relational database is perhaps the highest-value single technical skill for non-developers in the modern economy. Then Python basics — enough to automate file handling, clean data, and interact with APIs. This three-skill stack, achievable with roughly 200–400 hours of deliberate practice, transforms what a knowledge worker can do. Tools like Codecademy, Khan Academy, and fast.ai have made this pathway more accessible than it has ever been.
**For developers working with non-developer stakeholders:**
The practical implication is that you should actively encourage and support technical literacy in your non-developer colleagues. This means being willing to do code reviews for non-developers' scripts, being patient when product managers ask technical questions that reveal limited understanding, and explicitly welcoming the effort rather than using jargon as a moat. The collaboration dividend is real, and you personally benefit from working with technically literate colleagues.
**For organizations and engineering leaders:**
The evidence suggests that investing in coding literacy across the organization is a genuine competitive advantage. This does not mean training everyone to be a developer — it means investing in SQL workshops for analysts, Python basics for operations teams, and API literacy for product managers. Companies like Airbnb, Shopify, and Spotify have programs along these lines, and the productivity returns are well-documented in engineering and business literature.
**For curriculum designers and educators:**
The evidence base most strongly supports embedded, problem-focused computational thinking education rather than isolated coding instruction. Teaching children to think algorithmically through mathematics, science, and arts — as Finland does — yields stronger long-term outcomes than teaching them to recite Python syntax in a standalone class. The goal is developing a mental model of computation, not building a technical workforce prematurely.
---
SECTION 11: THE PHILOSOPHICAL DIMENSION
There is a dimension to this debate that rarely gets discussed in tech circles but deserves acknowledgment.
The argument that everyone should have some understanding of the systems that shape their lives is not primarily economic. It is a civic argument. We live in a world where algorithmic systems make decisions about credit scores, parole recommendations, hiring, medical diagnoses, content recommendation, and the price of insurance. These are not abstract concerns. They affect the distribution of life chances in democratic societies.
A citizenry that has no understanding of how algorithms work, how data can encode bias, how recommendation systems shape political discourse, and how software can fail in systematic ways is not fully equipped to participate in democratic self-governance in the 21st century. This is a more defensible version of the "everyone should learn to code" argument than most of its proponents articulate — and it is one that should resonate with any developer who thinks seriously about the social implications of the work they do.
Cathy O'Neil's "Weapons of Math Destruction" (2016) documented how algorithms operating on flawed data could systematically disadvantage the poor and vulnerable in ways that were invisible to those affected because they lacked the technical literacy to understand what was happening to them. Virginia Eubanks' "Automating Inequality" (2018) made similar arguments in the context of social services automation. Both books conclude, implicitly or explicitly, that digital literacy — including basic computational thinking — is a prerequisite for informed democratic participation in the current technological environment.
---
SECTION 12: THE VERDICT — A NUANCED POSITION
After examining the evidence, the case studies, the economic data, the computational thinking research, and the legitimate criticisms, a defensible position emerges.
Not everyone should become a professional software developer. The time and cognitive investment required to do that well is simply not justified for most career paths, and the claim that it is represents either deliberate oversimplification or a category error about what coding actually requires.
But: a very large proportion of knowledge workers in the modern economy — probably the majority of them — would benefit materially and professionally from developing basic computational thinking skills and practical coding ability at the level of SQL, Python scripting, and data manipulation. The evidence for this is strong, consistent across multiple study types and populations, and increasingly urgent as AI tools make basic coding more accessible while simultaneously demanding better technical intuition to use effectively.
And: all citizens in democratic societies that are governed by algorithmic systems have a stake in developing enough digital literacy to participate meaningfully in decisions about how those systems are designed, deployed, and regulated. This is not primarily a career argument. It is a political one.
The "everyone should learn to code" slogan is imprecise. But the underlying argument — that computational thinking, digital literacy, and basic coding ability are foundational skills for the modern world, not specialized professional credentials — is well-supported by evidence and deserves to be taken seriously by everyone, including the professional developers who most reflexively dismiss it.
---
SOURCES & REFERENCES
---
**Bureau of Labor Statistics — Software Developers, Quality Assurance Analysts, and Testers**
[https://www.bls.gov/ooh/computer-and-information-technology/software-developers.htm](https://www.bls.gov/ooh/computer-and-information-technology/software-developers.htm)
*Key data: "Employment of software developers, quality assurance analysts, and testers is projected to grow 25 percent from 2022 to 2032, much faster than the average for all occupations. The median annual wage for software developers was $127,260 in May 2023."*
---
**Lightcast (formerly Burning Glass Technologies) — "The Coding Edge" (2016)**
[https://lightcast.io/resources/research/the-coding-edge](https://lightcast.io/resources/research/the-coding-edge)
*Key data: "Jobs requiring coding skills pay an average of $22,000 more per year than positions that do not require coding in the same occupational category."*
---
**World Economic Forum — "Future of Jobs Report 2023"**
[https://www.weforum.org/reports/the-future-of-jobs-report-2023/](https://www.weforum.org/reports/the-future-of-jobs-report-2023/)
*Key data: "65% of employers expect AI and big data proficiency to increase in importance. Technology literacy and programming are among the top skills expected to increase in demand through 2027."*
---
**McKinsey Global Institute — "Skill Shift: Automation and the Future of the Workforce" (2018)**
[https://www.mckinsey.com/featured-insights/future-of-work/skill-shift-automation-and-the-future-of-the-workforce](https://www.mckinsey.com/featured-insights/future-of-work/skill-shift-automation-and-the-future-of-the-workforce)
*Key data: "By 2030, as many as 375 million workers globally may need to switch occupational categories as digitization, automation, and advances in artificial intelligence disrupt the world of work."*
---
**Stack Overflow Developer Survey 2023**
[https://survey.stackoverflow.co/2023/](https://survey.stackoverflow.co/2023/)
*Key data: "The median salary for a developer in the United States is $150,000. Over 90,000 developers participated, making it the largest survey of software developers globally."*
---
**GitHub — "State of the Octoverse 2023"**
[https://github.blog/2023-11-08-the-state-of-open-source-and-ai/](https://github.blog/2023-11-08-the-state-of-open-source-and-ai/)
*Key data: "GitHub surpassed 100 million developers globally in 2023, with the fastest-growing segments including scientists, educators, and domain experts building tools for their specific fields."*
---
**Jeannette Wing — "Computational Thinking" (Communications of the ACM, 2006)**
[https://dl.acm.org/doi/10.1145/1118178.1118215](https://dl.acm.org/doi/10.1145/1118178.1118215)
*Key quote: "Computational thinking involves solving problems, designing systems, and understanding human behavior, by drawing on the concepts fundamental to computer science."*
---
**Computers & Education — Meta-analysis on coding education for non-programmers (2019)**
[https://www.sciencedirect.com/journal/computers-and-education](https://www.sciencedirect.com/journal/computers-and-education)
*Key data: Meta-analysis of 105 studies found statistically significant improvements in problem-solving ability and logical reasoning after coding instruction for non-professional programmers.*
---
**UK Department for Education — Computing in the National Curriculum**
[https://www.gov.uk/national-curriculum/other-compulsory-subjects](https://www.gov.uk/national-curriculum/other-compulsory-subjects)
*Key data: "Computing became a statutory subject in September 2014, making England one of the first countries to mandate computing education from ages 5 through 16."*
---
**Royal Society — "After the Reboot: Computing Education in UK Schools" (2017)**
[https://royalsociety.org/topics-policy/projects/computing-education/](https://royalsociety.org/topics-policy/projects/computing-education/)
*Key quote: Early evidence suggests students who engage with programming concepts show improved abstract reasoning and stronger problem decomposition skills.*
---
**Estonian Education Ministry — ProgeTiiger Initiative**
[https://www.hm.ee/en/activities/educational-technology/progetiiger](https://www.hm.ee/en/activities/educational-technology/progetiiger)
*Key data: Estonia introduced programming education in 2012 and by 2021 ranked in the top tier of European nations for digital skills.*
---
**MIT Sloan Management Review — "Democratizing Data and Analytics" (2022)**
[https://sloanreview.mit.edu/](https://sloanreview.mit.edu/)
*Key data: "Companies with democratized data and analytics capabilities made decisions 2.4x faster and reported significantly higher employee satisfaction."*
---
**GitHub — "Productivity Research: Quantifying GitHub Copilot's Impact on Developer Speed and Code Quality" (2023)**
[https://github.blog/2023-06-27-the-economic-impact-of-the-ai-powered-developer-lifecycle-and-lessons-from-github-copilot/](https
Comments
Post a Comment