You missed the point. Here’s the same point without the sarcasm:
You need to prove your hypothesis before finding a solution.
Why?
Let’s say my goal is to guide objects into a hole.
I choose a red ball. I drop it into the system. It does not make it into the hole.
I pick up an identical red ball and drop it into the system again. It also does not make it into the hole. We have an issue: objects are not making it into the hole.
What hypotheses can we make?
-
Red objects are not making it into the hole.
-
Round objects are not making it into the hole.
-
No objects are making it into the hole.
The next step isn’t to pick one and fix the perceived issue. All of these hypotheses are supported by the evidence thus observed. If you spend time and effort building a red item detector that guides things onto the ramp but the issue isn’t that it’s red, you haven’t fixed the root cause. You need to find out if your hypothesis is right or wrong.
Drop a red cube into the system. Drop a green ball into the system. Drop a bigger item into the system. Drop a smaller item into the system. Drop many different balls at the same time into the system.
Improve your hypothesis. If red cube makes it into the system, hypotheses 1 and 3 are wrong… etc etc
Correlation does not imply causation. Fix the cause not the symptom.
They have enough skill to be dangerous, not enough skill to work on actual products.