The Manager's Guide – #75

Mar 25, 2024

You might have noticed the pause in newsletter issues - I was on a nice vacation in Italy, but now I’m back, so full speed ahead :)

How hard is it to cheat in technical interviews with ChatGPT? We ran an experiment.

Finally! This was long due. :)

🕵️‍♂️ Three types of questions were used: Verbatim LeetCode, Modified LeetCode, and Custom questions.
📈 Verbatim questions saw a higher pass rate, suggesting ChatGPT can easily solve familiar problems.
🔍 Custom questions proved challenging for ChatGPT, resulting in a significantly lower pass rate.
🚫 Surprisingly, no instances of cheating were detected by interviewers during the experiment.
🤔 Interviewers remained confident in their assessments despite the cheating, suggesting a potential blind spot.
📚 Highlight on the necessity for companies to innovate their interview questions to prevent cheating.

Personal take: the only thing this highlights (that we already knew) is that these types of interviews are mostly worthless.

No summary, go and read the level definitions :)

🚀 Fast deployment and thorough testing in software development are often seen as opposing goals, but using feature flags can achieve both by separating the deployment of code from its activation for users.
🧩 Adding a new dimension to decision-making can transform apparent trade-offs into opportunities where everyone's needs are met without compromise.
🔄 Examples across different contexts, like project budgets, investment strategies, and security measures, show how introducing an additional dimension can lead to better outcomes.
🤔 The process of adding dimensions to trade-offs involves believing in the possibility of improvement, understanding stakeholder needs in detail, exploring various dimensions, and consulting with peers.
🌟 Late in one's career, the ability to identify and apply new dimensions to solve trade-offs becomes a valuable skill, highlighting continuous learning and adaptation.

🐍 The Cobra Effect illustrates how solutions to problems can unintentionally exacerbate those problems.
🐭 Similar unintended consequences have arisen from incentives to control rat populations, reduce greenhouse gases, and decrease narcotic production.
🦏 In some cases, seemingly negative incentives, like legalizing the hunting of endangered species, can lead to positive outcomes, such as increased animal populations.
🐅 Keeping animals in zoos and aquariums, while controversial, has raised public interest in wildlife conservation.
💉 Safe injection sites and sex education programs have successfully mitigated negative societal issues.
📱 Unlimited data plans often result in minimal service levels as carriers aim to prevent customer loss rather than enhance user experience.
💬 Direct feedback initiatives can disrupt workplace relationships and efficiency, highlighting the nuanced role of management.
📋 Overly detailed job descriptions can limit employee initiative, underscoring the importance of job discretion and adaptability.
🔍 Cross-team calibration without deep understanding can lead to a superficial evaluation of value, emphasizing the need for meaningful impact recognition.
✨ Setting incentives is a powerful tool that requires careful consideration to avoid unintended negative outcomes.

Kind of kills the original setup of the rubber duck, still useful though.

🦆 Rubber duck debugging involves explaining a problem out loud to discover solutions.
🗣️ A human “rubber duck” can offer more value by asking specific questions to aid problem-solving.
🕵️‍♂️ Question 1: “How did you first start investigating this?” helps regain lost perspective and refocus.
🔍 Question 2: “What observations have you made?” encourages a review of all facts, aiding in hypothesis formation or adjustment.
🔄 Question 3: “If your hypothesis were wrong, how could we disprove it?” helps escape tunnel vision by considering alternative explanations.
🧠 These questions are rooted in hypothetico-deductive reasoning, aiming for consistent problem-solving despite complexity.

The eternal question, this time answered by research at Microsoft.

🌟 Maintaining a positive work environment is the top attribute for great engineering managers.
🌱 Growing talent and enabling autonomy are key, highlighting the importance of supporting engineers' development and giving them freedom in their work.
🛠️ Being technically knowledgeable is valued, but less critical than creating a supportive environment.
📊 Research from Microsoft involved interviews and surveys within the organization to identify and rank manager attributes.
🔍 Female participants placed a higher importance on technical knowledge compared to male participants.
🤝 Managers value team cohesion more than engineers, emphasizing the social dynamics and psychological safety in teams.
🚀 Successful management involves building a supportive, autonomous environment that boosts overall team effectiveness and is highly valued by engineers.

I hope you liked it, and you’ve learned something — if you did, don’t forget to give a thumbs-up and share this issue with your friends and network.

See y’all next week 👋