Over the last three weeks, I interviewed five manager candidates for my team. The conversations were fascinating, where I wish I had more time to hear more of their stories. Here are my key takeaways:
- I was surprised that most candidates were directors, managing multiple teams. I heard before that Netflix engineering managers operate at a director capacity. I wonder if that translated into the sourcing target.
- I was glad to see racially diverse candidates: three females, one Hispanic and one Black. I have given positive feedback to my director about this.
- I had a hard time judging between a weak yes and a weak no on the scale of 1-4. Even when I wasn’t too excited about the candidate, I felt like they would do a decent job and gave 3. Should that be a strong no, given Netflix’s high-performance culture? I need more calibration on the scale.
I loved talking to managers from small startups to big corporations. The notes from this experience will come in handy when I return to management in the future.
Software Engineering 🌐
“GraphQL has two distinct type systems by design: nominal (for output type) and structural (for input types).” – Ivan Goncharov
I have been confused about GraphQL backward compatibility because whether GraphQL uses a nominal or structural type system was not clear from the spec. This comment on Github is the only source I found about GraphQL’s nominal output type. To be more specific, input types are structural, but individual fields are nominal.
Speeding up a Git monorepo at Dropbox with <200 lines of code – Dropbox
It is always eye-opening when the seemingly constant operations like “git status” start slowing down at scale. I never thought about git commands’ linear performance.
Spark Joy by Running Fewer Tests – Shopify
Here is another article on tackling challenges with regards to a monorepo’s scale. Metaprogramming and dynamic types make Ruby an ideal language for small teams. However, the language does not scale as the organization grows.
As a side note, I like their chosen metrics to measure success. I wonder if Shopify’s deployment quality stayed the same.
2020 Developer Survey – Stack Overflow
I am late to the party here; Stack Overflow released the survey result late May. Rust is the most beloved language, and Windows is the most used development OS (surprise!). There is a gender divide as well: men are asking for Stack Overflow dark mode when women are talking about its toxic culture.
By the way, if you live in a non-English speaking country, I wouldn’t pay too much attention to it. The survey skews heavily towards the western world and the English speaking population.
Set an Aspirational Hourly Rate – Naval Ravikant
Since I started working, I tried to make tradeoffs in hourly rates, too. As a result, I treated my out of office time as “free” because none was going to pay me for it. So Naval’s point, “no one is going to value you more than you value you,” or instead, I should value myself more than the market does, is a significant shift for me.
Solving online events – Benedict Evans
An excellent dissection of offline conferences (content, networking, meetings) and what is easy to do online and what is hard. Will networking be accomplished online? More importantly, will it be less biased by where you are and how you look? I hope so.
Amazon’s New Competitive Advantage: Putting Its Own Products First – ProPublica
Amazon, like any other retail distributors, has its private brands. That’s why some argue this isn’t an anti-trust issue. But Amazon’s listing page isn’t just a dumb shelf at Walmart. It’s driven by a recommendation engine that consumers trust to produce the best results for them. There is potential consumer harm here when the recommendation pushes more of Amazon’s products.