First Principles Thinking

This week, I've revisited Sunil Pai's article "the senior engineer death spiral" multiple times. It resonates with me because I believe that almost every senior engineer has experienced some form of being stuck. As one gains expertise in their field, the accumulated experience can sometimes hinder the ability to approach problems differently.

Pai stresses focusing on momentum rather than outcomes. When faced with obstacles, I find it helpful to break down the work into the smallest achievable task. Accomplishing something, even a small task, often provides clarity on the next steps. Reflecting on the qualities of great senior engineers, I consistently arrive at the same conclusion: they possess an intuitive understanding of what needs to be done.

This intuition is a habit that sets them apart. Some of the best engineers I've collaborated with have backgrounds in customer support, services, or have self-taught themselves to code. Others began as designers or entrepreneurs. Regardless of their unique paths into engineering, they all share a common trait: they think from first principles.

They question the purpose behind building something and how it benefits the users. This ability to link codebase activities to real-world implications aids in maintaining simplicity. The concept of "agentic development" is a topic of discussion among friends and colleagues. Those who seem to thrive in this environment typically think from first principles.

This mindset allows them to embrace the "simulation switch-up" by temporarily setting aside their existing knowledge when working with agents. They experiment before assuming old constraints still apply. This is the challenging aspect for me. It's difficult to put aside one's experience, learned lessons, and current beliefs to view the problem anew.

However, stepping back and focusing on the core question - what we're truly trying to achieve, why it matters, and how the elements connect - often reveals more solutions than initially anticipated. There has been much debate about the reality of AI and its exaggerated claims. If we approach AI with a fresh perspective, setting aside our past experiences, there is much to admire and reconsider.

Pai's emphasis on first principles thinking makes this approach feel intuitive. When we deeply understand the problem at hand, taking small steps, learning from them, and progressing becomes more natural. This approach enhances the flow state when utilizing agents, enabling faster learning loops and momentum driven by profound understanding. This is the essence of the new flow state for me.