Dragan Stepanović

Trying to do more of Continuous Integration by trying to continuously integrate is a suboptimal intervention point in the system (can be a monitoring point, though).

for minimizing (lead) time to recover from incidents and the way to achieve it, somehow we fail to realize either the need for minimizing the lead time to value for our customers or the way to achieve it.

I tend to reason about the design of the code and system architecture by thinking about:

you’re working on something else than the person trying to interrupt you.

(some call this tech debt) piling up in the codebase because latency in the async code review process signals a high cost of review which in turn hinders continuous refactoring.

can only make a team slower if people need to integrate often. And some teams don’t realize how often they actually need to integrate.

through co-creation if a culture is not optimized for T-shapeness and nurtures learning.
Otherwise, you get entrenched siloing with unbalanced capacities that very often lead to blockers and pulling in more stuff.

Without the ability to learn and experiment faster in order to inform your product decisions, you’re mostly left with relying on the opinions of people inside the company.

Instant feedback loop guarantees that participating actors won’t have time to start anything else while waiting for response/feedback.

but a feature envied, tightly coupled, Asking instead of Telling production code.