engineering

Bilingual

You don't have to be smart if you can think clearly
2 months ago
- Smart engineers rely on quick intuition but struggle when problems aren't instantly solvable.
- Strong engineers use methodical processes and can think clearly even at slow speeds.
- The key to clear thinking is focusing on invariants—true beliefs about the system.
- Thinking slowly and clearly is often unsatisfying but yields high value.
- Engineers who can think clearly under uncertainty are extremely effective.
When the crisis comes - Anil Dash
2 months ago
- Crises can be opportunities for improvement and innovation, not just paralyzing events.
- Effective crisis response requires expertise, compassion, and a structured approach rather than guesswork.
- The book 'Crisis Engineering' by Marina Nitze, Mikey Dickerson, and Matthew Weaver documents proven methods from real-world crises.
- Organizations can plan for crises as ordinary events, building confidence and readiness in their teams.
an engineer's perspective on hiring
2 months ago
- Most companies fail at hiring by wasting time with lengthy interviews, chasing trendy skills, and treating all programmers as interchangeable.
- Good interviews should differentiate skills, reflect actual job duties, think long term, be time efficient, respectful, and assess 'taste' in engineering.
- Common methods like live coding, take-home tests, and architecture design often fail on key criteria like differentiation, applicability, or respect.
- Code review interviews reverse time asymmetry, allow assessing collaboration and taste, and are highly efficient and applicable.
- A better process combines code review with live discussion of work samples, plus a meeting with the future manager.
- Measuring interview effectiveness can include red-teaming with existing employees and regret analysis of rejected candidates.
Designing bug-proof engines
2 months ago
- Two strategies for designing bug-proof engines: make it impossible for bugs to get inside, or design it to tolerate bugs.
- Examples include Saturn V (perfectionist) vs Falcon 9 (tolerant), C (manual memory) vs Java (garbage collection), and storyboarding vs sloppy draft in writing.
- Engineers often assume failure is unacceptable; always question requirements and consider both ends of the solution spectrum.
- Bimodal systems have a perfectionist happy path and a sweeper sad path; often cheaper to just use the sad path to reduce complexity.
A Year of Writing for Myself (That 230,000+ People Read)
2 months ago
- Started a blog a year ago to organize thoughts on management and engineering with modest expectations.
- Received over 230,000 unique visitors in the first year, much more than anticipated.
- Viral posts included 'The Hidden Cost of AI Coding' and 'Stop Avoiding Politics', both hitting #1 on Hacker News.
- Favorite personal posts highlighted topics like seniority, strengths vs weaknesses, influential CS papers, and workplace politics.
- Realized the blog fills a gap in content about the human side of engineering, soft skills, and interpersonal dynamics.
- Writes mostly for personal organization but is grateful for the positive impact on readers.
Creating Your Own Opportunities
2 months ago
- Engineers often complain about not getting growth opportunities from managers, but complaining rarely helps.
- Even the best manager has limits; sometimes they have boring but necessary projects.
- Create your own opportunities instead of waiting for them; example of turning a simple UI task into a reusable design system library.
- Identify leverage points where a little extra effort creates much more impact.
- Doing only your assigned job rarely leads to promotion; going beyond is noticed.
- Managers may not have bandwidth or see gaps; your daily work gives you a unique vantage point.
- Do extra work for yourself (skills, reputation), not just for the company—choose high-leverage work within existing hours.
- Not all gaps are worth filling; good opportunities are visible, help others, and compound over time.
- First complete your actual responsibilities, then look for adjacent improvements no one owns.
- Bring ideas to your manager; they are often thrilled when you propose solutions.
- This advice assumes a decent workplace; if your effort is exploited, this won't help.
98% Isn't Much
3 months ago
- 98% success rate is context-dependent: excellent for exceptional achievements but unacceptable for basic expectations.
- In web development, 98% support for a feature can exclude a significant portion of users, especially if audience demographics differ from the general population.
- Robust engineering requires graceful handling of edge cases; lack of degradation for unsupported features means 98% isn't truly 'widely supported'.

engineering

You don't have to be smart if you can think clearly

When the crisis comes - Anil Dash

an engineer's perspective on hiring

Designing bug-proof engines

A Year of Writing for Myself (That 230,000+ People Read)

Creating Your Own Opportunities

98% Isn't Much