But Google (and Facebook, and probably some other companies) don't have reasonable processes for disputing or resolving these situations.
Some have said that we should consider Google's challenge: lots of users/activities that need to be monitored and policed. The assumption is that Google could not afford to do this "reasonably" with humans instead of automated systems because the volume is high.
But Google certainly could hire and train humans to follow a process for reviewing and assisting in resolving these cases. They don't. It is doubtful that they cannot afford to do this; I haven't checked their annual report lately, but I'm guessing they still have a healthy profit.
In the unlikely event that involving more humans would be too expensive, then Google should raise their prices (or stop giving so much away for free).
To summarize, there is no excuse for Google to operate this way. They do because they can, and because the damage still falls into the "acceptable losses" column.
They didn't have exactly what you wanted so provided a workaround that would solve the problem.
I have never heard anyone say that the AWS toolset was anything but "an amalgam of individual projects developed separately." It is obvious from their UI that the different tools are run by different teams that have very different opinions on how things should be done. Just look at the various iterations of deployment management. ECS vs Lambda vs EKS vs classic EC2. All the UIs have different design standards and assumptions. It has gotten better over the years, but the AWS org chart is still peaking through the UI.
GCP is not much better. At least they had the advantage of starting later in the market cycle. They were able to see what worked and what didn't work at AWS and build a bit cleaner.
In the end we are talking about B2B systems targeting power user engineers. The control surfaces need to be powerful first, and easy to use is a distant second or third consideration.