The End of Technical Interviews: Why Our Hiring Process Is Broken and What Might Replace It

In a provocative post that's been 35 years in the making, Steve Yegge declares the technical interview process on its last legs. Having conducted technical interviews for decades and worked to improve the process at both Amazon and Google, Yegge brings insider perspective to what he calls an 'embarrassingly busted' system that hasn't meaningfully changed in fifty years.

The Broken System

At Google, internal studies revealed depressing truths about technical interviews: interviewers barely agreed with each other, with the same candidate routinely receiving a 'strong hire' from one interviewer and a flat 'no' from another. Once hired, interview scores told you next to nothing about how candidates would actually perform on the job. Some of Google's star performers failed their interviews four or five times before finally getting in after more than two years.

The most telling evidence came when Google's Hiring Committee reviewed anonymized interview packets and voted not to hire about two-thirds of them—only to discover these were their own interview packets. This moment of clarity revealed how utterly broken the process was, yet the company never fixed it.

"Interviewing, it turns out, is a big game of darts," Yegge writes. "A 'do I like you' dating round."

Current Band-Aids That Don't Work

Major companies have implemented various band-aids to compensate for the flawed process:

• Amazon's Bar Raisers and Microsoft's As-Appropriate roles that veto hiring decisions
• Hiring Committees that override interview teams
• Multiple interview loops in a single day

"These are tacit acknowledgments that you can't trust your interview teams to make good hiring decisions," Yegge notes. "Which even more broadly suggests that if every single interview loop needs a babysitter, then it is a flawed process."

The Signal Problem

Talent assessment is fundamentally a problem of signal collection. Current methods provide woefully inadequate data:

• Résumés: "A piece of paper with lies written all over it" with terrible signal-to-noise ratio, exacerbated by AI-assisted writing
• Phone screens: Better than résumés but still limited
• Work samples and OSS contributions: Useful tie-breakers but don't involve working directly with the candidate
• In-person interviews: Infamously unlike real-world work

Today's standard hiring process collects only a few hours of signal to make decisions that could last years.

Provisional Employment: The Current Best Solution

The gold standard for talent assessment, when available, is provisional employment:

• Internships (typically 3 months)
• Co-ops (like Geoworks' 6-month program)
• Contract-to-hire arrangements

These provide dramatically stronger signal than interviews because they assess actual work performance in real contexts. Geoworks' 6-month co-op was so effective that Amazon later acquired their Seattle office, which became "by far the strongest group of people they ever brought in." Many of these engineers went on to do exceptional work at Amazon and Google.

The University of Waterloo takes this further, requiring Computer Science students to complete six internships—roughly two years of real-world work experience before graduation.

Why Provisional Employment Isn't More Common

Historically, engineers had abundant options and wouldn't risk uncertain employment. Additionally, U.S. employment law makes it difficult to fire underperforming permanent employees, creating high-stakes hiring decisions. Provisional employment serves as a workaround to this legal constraint.

The Changing Landscape

Several forces are converging to make traditional interviews obsolete:

• AI is making résumés less reliable
• The nature of work is changing rapidly, making traditional assessment methods obsolete
• The system is under strain from AI cheating, fraudulent interviews, and résumé floods

"We no longer know what job roles to post, what questions to ask, nor how to evaluate candidates properly using any of our existing processes," Yegge writes.

Proposed Solutions

Yegge suggests several radical alternatives to traditional interviews:

The Campfire Model

"Pull up a log, build something together, see how it feels." This approach brings candidates in for longer stints to work on real projects, generating better signal than standard interviews. Companies like those in the San Francisco tech scene are already experimenting with this model.

Make Work Count Twice

Carve real pieces of work from your project backlog and have candidates complete them. The work benefits the company while providing candidates with portable credentials they can keep regardless of hiring decisions. This transforms "six hours of free work to get rejected" into "a six-hour investment in a permanent small reputation bump."

Make Interviewing a Profit Center

When done correctly, this approach flips interviewing from a cost center to a profit center. Companies "hand out stamps" (assessments) that have lasting value to candidates, creating a reserve of proven workers who can be called upon when needed.

Portable Assessment Systems

The key innovation needed is assessment systems that candidates can carry with them, similar to how Uber and Grab track driver performance but with the ability to take those credentials elsewhere.

Conclusion

Yegge doesn't offer a single silver bullet but argues that we need to fundamentally reimagine talent assessment. "As long as it's not more of the same," he writes. "I just wanted to move the Overton window a bit."

The technical interview, he predicts, will soon become "a cute historical footnote like phrenology." In its place will emerge systems that evaluate actual work performance rather than simulating it through artificial exercises.

The future of hiring, Yegge suggests, isn't about asking better interview questions but about creating systems where candidates can demonstrate their value through real work while building portable credentials that benefit them regardless of hiring outcomes.