Real talk: the technical interview is broken

Here's the tell: at least five partner companies told Code2040 they're starting from scratch on technical interviews. Not tweaking the process. Starting over. They know what Code2040 suspected: current interviewing provides mediocre, biased results.

The problem isn't just broken, it's non-existent. There's no industry or even company consensus about what a useful technical interview looks like. No universal understanding of what must be tested, why it's part of the conversation, or how to fairly evaluate results. Meanwhile, a cottage industry has emerged that looks uncomfortably like SAT prep—candidates spending thousands of dollars learning cultural norms to get a desk at a tech company. And research tells us the SAT doesn't predict college or career outcomes.

Code2040 convened a think tank with GitHub, Slack, Medium, Lyft and others. Here's what they found: interviews have devolved into game shows where the prize is an offer letter. Three unproductive patterns dominate: obsession with brand names (elite schools, well-known companies) that disadvantages people marginalized by race, class or gender; algorithm quizzes unrelated to day-to-day work; and artificial whiteboard tests that trigger stereotype threat.

The fix: evaluate projects candidates built on their own terms and ask about design choices. Measure ability to learn, not rote CS fundamentals. Code2040's intern challenge lets students code against a simple API in any stack, providing real-time feedback. The challenge assumes technical skill isn't fixed—that googling, persistence and experimentation matter more than memorized algorithms. When people practices aren't explicit, they disproportionately impact marginalized communities. Get explicit.