I am sick of LeetCode-style interviews

Perceived Problems with LeetCode‑style Interviews

• Many see them as artificial puzzles disconnected from day‑to‑day work, especially for web/product roles.
• Strong sense they select for memorization, test‑prep, and free time, not real problem‑solving or engineering judgment.
• Candidates describe freezing on simple questions under pressure; interview anxiety is seen as orthogonal to job performance.
• Senior engineers with long track records report failing or needing to “grind” to pass, which they experience as demeaning and age‑biased.
• Some see LC usage at low‑pay or non‑FAANG orgs as pure cargo‑cult: copying big tech without matching needs or compensation.

Arguments in Favor / Defense of LeetCode

• Defenders say it is:
  • A scalable, reasonably objective way to check basic coding and CS fundamentals.
  • Better than pure “vibes” / resume / school‑name screening.
  • A way to avoid false positives, which are seen as costlier than false negatives.
• Some report that simple coding screens (FizzBuzz, basic array/hash problems) reliably weed out many candidates who cannot code at all.

Fairness, Bias, and Access

• Critics argue LC systematically favors:
  • Younger candidates and those without family or caregiving obligations.
  • People with time and energy to grind months of problems.
• Others counter that any assessment requires prep; without tests, bias shifts back to pedigree and networking.
• Several note LC can both mask and enable bias: same process, but easier questions and “help” for favored candidates, harsh treatment for disfavored ones.

Alternatives Proposed

• Work‑sample tests: small, realistic tasks (feature in an existing codebase, bug fix, code review, refactor) with clear rubrics.
• Pair‑programming on a relevant problem, or walking through past real work instead of puzzles.
• Short, basic coding plus deeper discussion of previous projects, architecture, and trade‑offs.
• “Fix this failing test” or “debug this broken system” style exercises seen as closer to real work.

Cheating, LLMs, and Take‑homes

• Take‑home tasks feel fairer to candidates but many hiring managers report rampant cheating (LLMs, second person on call, leaked question banks).
• Some argue that using LLMs is realistic—tools will be used on the job—so the bar should shift to code quality and follow‑up discussion.

Meta: Interviews Are Broken and Hard

• Broad agreement that interviewing is noisy, hard to standardize, and often poorly executed.
• Several conclude LC is “the least bad” current scalable tool, while others prefer to walk away from any company that uses it heavily.