Racing used to punish mediocrity
Racing used to punish mediocrity.
With a combustion engine, the car was a system you had to actively manage:
gear selection, power band, traction, cooling, failures you could feel coming if you were paying attention.
Speed came from judgment, not just bravery.
Electric racing changes the shape of the problem.
You can be “good enough” and still put down shocking lap times because torque, control systems, and consistency remove a lot of the old penalties. The floor rises fast.
But the ceiling moves even faster.
The serious advantage shifts to people who understand the system behind the system:
energy management over a stint, thermal limits, regen strategy, software tuning, the places where the car will quietly protect itself until it suddenly can’t.
Coding is going through the same transition.
LLM code gen makes it possible for a mediocre engineer to ship something that looks fast. Working code appears. Demos land. Velocity feels real.
Until you hit the parts that used to separate builders from typists:
architecture under changing requirements, failure modes, observability, security boundaries, performance cliffs, “it works” vs “it keeps working.”
The new gap is not who can write code.
It’s who can drive a system they didn’t fully write, at the edge, without crashing it in production.
The floor is rising.
If you cannot explain why the generated code works, you do not own the speed.