Comparison
Cargo cultvsFirst principles
Cargo cult
the team adopted the whole process from a conference talk without the constraints that made it work there.
Copying the visible practices of a successful team without the conditions that made those practices effective. It shows up as microservices at ten engineers, a review process from a company of thousands, or a testing pyramid nobody can explain. The diagnostic question is why this works here, and a copied answer that cannot survive it is ritual rather than engineering.
Full entry →First principles
you stopped comparing frameworks and worked out what the system actually has to do, in numbers.
Reasoning from the underlying facts and constraints rather than by analogy to what others do. It is the move that ends an argument stuck on precedent — how many requests, how much data, what latency is actually required — and it usually shrinks the problem dramatically. It is expensive, which is why it belongs on decisions that are one-way, and why using it on everything is its own dysfunction.
Full entry →