How good of a developer are you?
summary
Certifications and whiteboard tests predict knowledge but not real-world coding skill; meta‑studies show low predictive power (one study with 3,500 students) and correlations like 0.16–0.23 for degree and early performance. Research repurposes the Dunning‑Kruger self/peer rating plus time‑bounded, code‑writing tasks, automated quality scoring and a Rasch model to norm results. A funded platform called CS collects control groups, compares scores, and correlates higher skill with fewer failed projects (about a 10% difference), so the method actually works.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:39Measuring developer skills
- 3:02Hiring methods flaws
- 5:49Good developer research
- 7:15Education vs skill
- 9:21Knowledge vs skill
- 10:44Skill levels defined
- 13:09Skill and motivation
- 15:20Gunnar Berson’s research
- 18:50CS platform for skill
- 20:35Future of skill assessment
related talks
24mInterviews for developers, by developers
56mJ-Fall 2023: Roy Wasse & Roy van Rijn
26mSkill Based Economy: The Evolving Role of Certifications for Developers
33mCan AppSec Training Really make a smart developer
45mCan AppSec Training Really Make a Smarter Developer?
01h 08mBeginner's Mind, Expert's Mind
58mThe No-BS Guide to Future-Proof Your Career
51mYour worth as a developer: not what you do, but why