I think the problem is, ironically, trying to use programmers for this. Not only are most of them opposed to vibe-coding out of principle; they work in ivory towers of pure math and can’t seem to wrap their heads around the real world and real users being random; they also have surprisingly little respect for deterministic behavior, statistics, reproducibility, controlled environments, ACID, CAP theorem, you name it, they just don’t give a shit as long as they can play with code.
Should use QA people instead. Their entire work revolves around getting reproducible results; they don’t mind what tools they use to get there; they’re not burdened by preconcieved dellusions of grandeur about knowing how to program; and they understand that the real world can be unpredictable.
I think the problem is, ironically, trying to use programmers for this. Not only are most of them opposed to vibe-coding out of principle; they work in ivory towers of pure math and can’t seem to wrap their heads around the real world and real users being random; they also have surprisingly little respect for deterministic behavior, statistics, reproducibility, controlled environments, ACID, CAP theorem, you name it, they just don’t give a shit as long as they can play with code.
Should use QA people instead. Their entire work revolves around getting reproducible results; they don’t mind what tools they use to get there; they’re not burdened by preconcieved dellusions of grandeur about knowing how to program; and they understand that the real world can be unpredictable.