AI summary by TwinMind
Why “Learn to Code” Failed
PolyMatter
25m 26s
UC Berkeley Computer Science Explosion (2011-2021)
- CS graduates increased by 1,116% between 2011-2021
- At current trend, all 30,000 undergrads would study CS in 7 years
- By 2029, would produce more CS graduates than California's population
- Fewer software developers employed in US today than 6 years ago
University CS Dominance Statistics
- Berkeley not most CS-obsessed university
- MIT: 40% of students study same subject
- MIT 2023 graduates: 7 chemistry majors vs 266 CS&E majors
- Electrical Engineering & CS twice as popular as CS&E alone
- Berkeley, MIT, Cornell created dedicated computing colleges in 2023
- First new school since Eisenhower presidency
"Learn to Code" Cultural Movement Origins
- iPhone released 2007, followed by Uber, Airbnb, Instagram
- Cultural shift from Wall Street to Bay Area tech
- Shows like Social Network, Silicon Valley mythologized tech founders
- Obama called coding "ticket to middle class"
- White House promoted Computer Science Education Week, Hour of Code
- Warned of 1 million STEM graduate shortfall by 2022
- Bloomberg announced learning to code in 2012 (no evidence he did)
Political Appeal of "Learn to Code"
- Republicans: aligned with vocational education vision
- Corporations: steady supply of pre-trained labor at taxpayer expense
- National security: strengthen America vs China competition
- Diversity advocates: expanding opportunity narrative
- Democrats: empowering solution during Great Recession
- Addressed globalization concerns: layoffs, automation, outsourcing
K-12 Implementation
- Nearly 15,000 US high schools offer CS classes
- 37% of middle schools offer CS classes
- 11 states require CS class for graduation
- Red and blue states adopted Python/JavaScript curricula
University Infrastructure Crisis
- Total undergraduate CS students tripled in 10 years
- PhD students becoming professors stayed constant
- PhD stipend: $40,000/year vs $200,000+ at Amazon/Netflix
- Result: impersonal factory-like CS departments
- Class sizes: 400-600 students in introductory courses
- Undergrads deputized as TAs
- Universities require reapplication to CS programs after 1-2 years
Lottery System Examples
- Swarthmore, University of Maryland, UCSD use lottery systems
- Students pay $90,000+/year at Swarthmore, can be denied major by random number
- Students graduate with debt, minimal professor contact
- Inadequate job market preparation
- CS curricula focus on algorithms/theory vs marketable syntax/languages
- Graduates dabble in multiple languages but master none
Coding Bootcamp Rise
- Solution to university problems: 6-figure Amazon jobs
- Cost: $10,000-$30,000 for 12 weeks or less
- Stripped general education, taught to tech interview tests
- Peak: hundreds of bootcamps graduated nearly half as many as 4-year programs
- Revenue: north of $200 million/year
Bootcamp Challenges
- Students' performance affected bootcamp reputation
- Silicon Valley small world: most target same 5 companies
- Hack Reactor: 3% acceptance rate (same as Harvard)
- Teacher shortage: competed with big tech salaries
- Hired recent graduates who couldn't find work elsewhere
- Pressure to raise tuition as costs increased
University-Bootcamp Partnership Model
- 2011 Department of Education loophole created
- Universities could pay third parties if part of larger service bundle
- Online Program Manager (OPM) partnerships emerged
- OPMs created curriculum, hired teachers, recruited students
- Universities provided name recognition for 40% cut
- Bootcamps gained access to federal student loans through accreditation
Tech Industry Boom Period
- Pandemic era: explosive tech growth
- Companies hoarded labor aggressively
- 2023 Wall Street Journal: Meta paid woman $190,000/year "to do almost nothing"
- Aggressive hiring rewarded by investors
- Competition for talent fierce
Tech Industry Crash
- 2023: nearly 500,000 tech workers laid off
- 2022: 250,000 laid off
- 2024: another 250,000 laid off
- Total: ~1 million layoffs (equivalent to early 2000s China shock manufacturing losses)
- Tech unemployment rate now higher than national average
- Berkeley professors report best students can't find jobs
- Recent graduates had offers rescinded
Bootcamp Industry Collapse
- 2U declared bankruptcy
- Launch Academy paused enrollment
- Dev Boot Camp (acquired by Kaplan) closed permanently
- Most bootcamps no longer exist
Economic Analysis
- Interest rates rose from ~0% to 5%+ in 2022
- Tech industry unusually exposed to borrowing costs
- Software requires massive upfront investment
- Oversupply of programmers made workers expendable
- Employers treat programmers as "spigot" - on/off demand vs long-term investment
Future Projections vs Reality
- Bureau of Labor Statistics: 11% growth in computer occupations by 2033
- White House predicted 1 million STEM shortfall by 2022
- Reality: fewer software developers than 6 years ago
- Current tech employment: ~1.9 million (2.3% of US labor force even if doubled)
- Programming couldn't absorb 7 million unemployed Americans
Fundamental Misconceptions
- 2012 article compared coding to reading/writing
- Belief everyone would need to program (electricians, teachers, insurance adjusters)
- 1,000% growth seemed reasonable under universal skill premise
- Reality: programming is specialized career among many
- Tech as volatile as other industries
- AI rise demonstrates continued volatility
Human Diversity Ignored
- Not everyone suited for/enjoys programming
- Office work, sitting all day, computer focus, teamwork, deadline pressure
- Single mothers, laid-off factory workers, 18-year-olds treated as interchangeable
- Bootcamps promised $150,000 salaries for all backgrounds
- Minimal support for non-traditional students
- Former contractors struggled vs peers with coding experience
- Students blamed themselves when told coding was "easy"
Real-World Outcomes
- Many graduated with $20,000 debt, 3 months lost wages
- Bootcamp resume line worked against some applicants
- Even successful graduates vulnerable to layoffs
- Amazon laid off 18,000 employees January 5, 2023
- "Learn to code" no more magic than "learn wind turbine service" or "learn occupational therapy"
- Occupational therapy demand: 60% growth expected
- Wind turbine service: 22% growth expected (both faster than software)
Alternative Approach
- Invest in non-industry-specific skills
- Problem-solving, critical thinking, adaptability
- Higher-level skills vs specific programming languages
- Programming languages come and go
- Process-focused learning vs outcome-focused
