I went to see a circus this weekend and was amazed by the performers’ extraordinary physical feats. On the way home, I thought about how I only learned to fully appreciate those feats and their talent and dedication to reach that level after I started training my body.
This experience made me think about how we often judge others’ work without fully understanding their challenges. For example, we might think that customer support is a simple job, but those who work in customer support know that it can be very demanding. Similarly, we might believe executives have an easy job, but they often have to make difficult decisions despite uncertainties.
The truth is, we can’t fully appreciate the work of others unless we walk in their shoes. Yet, it is impossible to try every role to build that context. That’s why being open-minded and learning from our colleagues is essential. By taking the time to understand their challenges, we can collaborate better.
Software Engineering ⚙️
TypeStat: Converts JavaScript to TypeScript and TypeScript to better TypeScript.
Once you start a TypeScript project with a less strict configuration, it becomes quite challenging and tedious to enable the stricter configuration (such as strictNullCheck or noImplicitAny). I’ve used Airbnb’s open-sourced project ts-migrate to enable those options, but the project could not fix the existing errors (and we had 20k+ errors!). This new project called TypeStat promises to fix the existing errors, and I can’t wait to try it.
Exploring GraphQL Clients: Apollo Client vs Relay vs URQL
When you adopt GraphQL, the client you choose will influence your architecture significantly (which I didn’t think about when I first used GraphQL). I’ve learned now that URQL is fit for the simplest use cases, Relay is for the most scaled and complex use cases, and Apollo Client is the middle ground between the two.
The question I am still thinking about is if there is a viable migration path from one library to another. If not, should you always choose Relay since it can solve the most complex use cases?
People ❤️
Meta updates RTO policy with stricter mandate
“Culture is the behavior you reward and punish,” and big tech companies are finally backing up their RTO policies with punishment. Meta announced that the employees’ compliance with RTO will be reflected in their performance evaluation, just as Google announced a few months ago.
Business 💰
HashiCorp Abandons Open Source for Business Source License
HashiCorp, the owner of a popular infrastructure management tool, Terraform, changed its license to prevent competitors from using it. HashiCorp is following other companies like MongoDB and Redis Labs, which abandoned open-source licenses for their business gains.
At this point, I wouldn’t build a company based on an open-source technology owned by another company without a contingency plan to maintain a fork.
Cruise: Pedestrian caused North Beach traffic jam, not Outside Lands
A day after CPUC allowed Cruise and Waymo to operate a paid service in San Francisco at all times, Cruise’s robotaxis malfunctioned and created a major traffic jam at a music festival. The crowd made navigating the streets challenging and degraded the cell service these robotaxis depend on. This incident shows that the safe rollout of robotaxis needs not just better machine learning but also contingency plans for extreme environments.
Nvidia GPU shortage is ‘top gossip’ of Silicon Valley
This article taught me two mind-blowing facts. First, a top-of-the-line GPU costs as much as $35k. Second, it takes over a few thousand of those GPUs to train the current generative AI models, which explains the GPU shortage.