The Last Technical Interview
Technical hiring, especially at large tech companies, is widely criticized as unreliable, time‑consuming, and biased toward people who are good at performing interviews rather than doing the job. Commenters weigh alternatives such as work-sample tests, short-term “provisional” employment, internships, and standardized exams, while highlighting practical barriers like legal risk, candidate time, AI-assisted cheating, and the difficulty of firing bad hires. Many argue that no process will be perfect, so companies should accept more risk, focus on real-world work and reputation signals, and treat hiring as an ongoing relationship rather than a one-off gatekeeping event.
Limits of Current Technical Interviews
- Many commenters agree interviews are statistically poor predictors of on‑the‑job performance: same candidate gets opposite outcomes, scores don’t correlate well with later success.
- Psychometrics concepts are invoked: interviews lack reliability (different interviewers disagree) and thus can’t have high validity (weak correlation with real ability).
- Big companies’ stated goal of minimizing “bad hires” is seen as sensible, but the tools they use (leetcode-style loops, opaque committees) are viewed as mostly random filters.
Alternative Assessment Models
- Strong support for multi‑month “provisional employment” / internships / co‑ops as the highest‑quality signal: real work in real conditions.
- Critics argue this is only practical for unemployed or financially secure candidates; few will quit a stable job for a speculative “maybe” offer.
- Some note that regular employment already functions like provisional employment via probation periods and easy firing (in some jurisdictions).
Work-Sample and Take-Home Tests
- Many see work-sample tests as the practical “gold standard”: closer to real work, more controlled, less stressful, and often more time‑efficient than full onsite loops.
- Objections: they can demand many unpaid hours, discriminate against people with limited free time, and are now easily solved or boosted via AI tools.
- Some companies counter this by: (a) strict time budgets, (b) in‑office work samples, (c) evaluating how candidates explain and extend AI‑assisted work.
Organizational and Managerial Factors
- Several argue the real failure is organizational: conflict‑averse, weak managers don’t fire poor performers, so hiring must over‑optimize against bad hires.
- “Performance management” systems are described as arbitrary; anecdotes show both promotions despite missed goals and exits despite met goals.
Candidate Experience and Fairness
- A long personal story highlights repeated big‑tech rejections despite strong skills and intense preparation, leading to bitterness and suspicion of age/diversity bias and luck.
- Many see current processes as “nerd revenge” or hazing: heavy leetcode prep, opaque culture screens, arbitrary difficulty variation between candidates.
- There’s concern that filtering (ATS, recruiters) is more broken than the technical interviews themselves, with huge variance in recruiter quality.
Standardized Tests and Professionalization
- Some advocate standardized cognitive or aptitude tests as cheaper, more reliable tools; others fear political, ethical, and class‑stratification issues.
- Comparisons to trades: multi‑year apprenticeships and formal licensing are suggested as a model, including talk of software “building codes” and professional engineer certification.
Reputation / Stamp Proposals
- The article’s idea of “stamps” from campfire/provisional stints gets mixed reactions.
- Supporters like a portable, positive‑signal history; critics see it as effectively a public ledger of rejections and an additional, employer‑skewed hoop.
Meta Reactions to the Article
- Some praise the piece as an honest “kitchen confidential” about FAANG hiring dysfunction.
- Others view it as detached from real constraints (family, mortgages, current jobs) or as setting up a commercial product/agenda.
- Several emphasize that interviewing is inherently inexact; the goal should be “good enough and humane,” not perfect.