This is where I have found review to be really tricky. I have reviewed so much unreadable code that I lose faith sometimes. More to the point, readability issue

This is where I have found review to be really tricky. I have reviewed so much unreadable code that I lose faith sometimes. More to the point, readability issues are often perceived as nitpicks. Eventually after a few rounds of readability issues being addressed I can tell what the actual code is doing. At that point the major suggestions and some rework comes out, but now the author feels burned, nitpicked, and now instead of being done "their perfectly functional" code needs more work and the whole thing is not done. Do this a few times and relationships begin to break down. Hence, I now try really hard to ignore readability until the end. I don't know if this partly just hostile developers, different understanding of code. Ie, is code for humans to read, or for machines to execute? Or if it's dealing with feature factory developers that just want to keep adding despite how brittle the code is becoming. I presume the problem is somewhere in between, but to another extent the normalcy of the absurd where it is normal to spend 5 minutes per line of code to understand it, it is normal to havk in features and do manual regression testing, etc..