Becoming a dungeon master for an interview

Original Article ↗ Hacker News Discussion ↗

Technical hiring is increasingly criticized for relying on LeetCode-style algorithm puzzles that select for test prep and trivia rather than day-to-day engineering ability. Commenters explore alternatives like “dungeon master” or simulation-based interviews that model real support or product work, highlight trade-offs such as role‑playing discomfort, bias, and unpaid take‑home tasks, and suggest more structured rubrics or project-based conversations instead. Underneath is a broader worry that companies are optimizing for easily measured proxies—speed on contrived problems, willingness to jump through hoops—at the expense of real-world skills, collaboration, and mutual fit.

Simulation / “Dungeon Master” Interview Approach

  • Many find scenario-based, DM-style interviews more engaging and closer to real work than puzzles.
  • Benefits cited: reveals troubleshooting style, prioritization, communication, and how candidates handle partial information.
  • Related ideas: in-basket tests, “consultant for a day,” mock support/debug sessions, toy versions of real systems.
  • Concerns:
    • Roleplaying can feel cringe or stressful; better to treat as hypothetical discussion than theatrical acting.
    • Strong bias risk if interviewer withholds key context, expects a hidden “correct” answer, or penalizes realistic behavior (e.g., using help menus).
    • Disadvantages candidates who struggle to think, talk, and roleplay under time pressure, even if they’re good at the job.

Transparency and Mutual Evaluation

  • Some describe processes where late-stage interviews explicitly surface the role’s downsides and “wart”s, even trying to talk candidates out of accepting.
  • Goal: no surprises; interview behavior matches day‑to‑day behavior; both sides gauge realistic fit.
  • Limits: can’t fully discuss proprietary tech, personnel issues, or sensitive assessments.

Leetcode / Algorithm Interviews

  • Strong criticism:
    • Viewed as hazing, Goodharted proxy, and “anti-pattern” that tests exam prep, willingness to comply, or conscientiousness more than job skills.
    • Heavily gamed; candidates memorize patterns; even some interviewers misunderstand their own questions.
    • Filters out experienced people with existing obligations and options who won’t devote weeks to practice.
  • Defenses:
    • High-paying companies need some standard, demonstrable bar; leetcode is one such bar.
    • Shows basic DS/algorithms knowledge and weeds out fast-talkers.
    • Some enjoy it as a puzzle-like hobby and credit it with helping them enter software.
  • Ongoing arms race: harder questions as candidates prep more, further distancing interviews from actual work.

Alternative Interview Philosophies

  • Emphasis on:
    • Discussing real past projects and architectures in depth.
    • Walking through actual backlog items or realistic design problems at sensible scale.
    • Short, pragmatic coding or design tasks; minimal gimmicks.
    • Standardized rubrics and panel review to reduce individual bias.
  • Many object to long, unpaid take-homes and multi-round “gimmick” processes; they treat them as red flags and proxies for company culture.