on
Starting With Failure
Despite being less talented than many of my classmates, I outscored them on the GRE with a 1590⁄1600. I beat or tied people who were demonstrably better suited for grad school than me, because I spent more hours memorizing words for a test. The GRE was a terrible measure of skill; it mostly measured monomania and resources.
Last week I failed my systems design interview at a FAANG company. I am most upset at myself and the choices I made, as those determined my outcomes. But I am second most upset at the system, because afterward I was clued into several of the expectations and practices for passing this interview format. Similar to the GRE, I learned that performance on this challenge is primarily a function of preparation, practice, access to someone to correct you, and being trained in the odd social dynamic of the test. That’s disappointing for several different reasons, not least among them that it is a test of navigating poorly-socialized material. (Easy answer: socialize the expectations better.)
Humility is a technologist’s virtue, so I am going to recover from this setback the hardest way: publicly. Hopefully, it will be of use to others facing similar challenges.
In the Situation, Task, Action, Result framework, the setup is simple:
Situation: I failed a significant test.
Task: I want to make this test a non-issue for the rest of my life.
Action: I will study and chronicle my answers over time, as they start from a bad place and hopefully progress.
I don’t know what the Result will be yet. But I do know what my stretch goals are.
Level 1: I indisputably nail these exams and remove them as barriers.
Level 2: If I am right that systems design interviews measure time and access instead of skill or capacity, I want to make it so straightforward for other people to pass that interviewers and interviewees acknowledge the social fiction of the test. Ideally, I want enough interviewees to pass with flying colors that interviewers need to recognize they are measuring time, not talent.
Hilariously overambitious Level 3: If I am right, and the systems design interview is eventually acknowledged as measuring capacity or skill, I want to see it dislodged from the interview pipeline and replaced with a test with better predictive signal.
I have no idea if this will work. But tomorrow is the first draft, with a writeup of an SDI problem and feedback from a very accommodating, more-talented friend.