Engineering Philosophy 2026
Engineering orgs run on a clear north star, access to truth, and a high learning rate.
Engineering leadership is about picking the right north star, expressing it well, repeating it often, and then working every day to remove barriers that prevent access to truth or limit the learning rate.
Rate of improvement matters more than starting point (bitter lesson).
from Tinker, but with an investor’s mindset. Control emotions when deciding where to spend or cut effort.
Architecture limits the possible rate of improvement. It determines which S-curves to ride and which others to push. Good architecture assembles components in a way that lets them benefit from each others’ strengths, while insulating them from each other’s weaknesses (minimizing tail risk).
For software engineering, org structure is the architecture of the machine that builds the machine. Bad org structure dilutes truth (the user!) and partitions the design space in ways that make the right solution inaccessible (Conway’s law). Learn to reorg fast and drama-free, so that you never have to get it perfectly right.
Problems get solved by the right people agreeing that there is a problem and then talking to each other. Org structure provides the networking. The finest grained org unit, the individual, has highest bandwidth and lowest latency. Higher bandwidth => more decisions; lower latency => faster re-alignment.
Lastly, a focused engineering org gets tunnel vision. There are always opportunities to achieve better outcomes for users now. A single owner can’t perform equally well on two time horizons. What would it take to deliver the outcome in half the time? What would it take to double the impact?