Compared to a software product, the scientists have build a product that contains many modules that all link together via complex maths. What is still missing

Compared to a software product, the scientists have build a product that contains many modules that all link together via complex maths. What is still missing is refactoring: making things simple again and removing unnecessary parts. While many of the modules have been tested separately, certain combinations may not work as proclaimed. I find this problem already with the physics of the sun. So many things are way off normal physics. I have seen models that do not exist anywhere else. Predicted values are 10^6 order off. These things need retesting and the modules likely need to be redesigned. But just stating that something might be off, already triggers many scientists. They see it as an attack on "their" science. This is clearly an attack on the messenger. Usually mixed with logical fallacies. So there is a real problem with the involved scientists as well. They do not want to see errors in their system, as it hurts their status. And there is the real problem with science. The ones involved do not want to admit that there might be something wrong. If I would tell programmers that something might be wrong with one combination of modules, they are (more) often happy to look into it. So scientists, be more like programmers.