Thanks for sharing your experience, much appreciated, Peter! I suppose the main purpose of the article is indeed to raise awareness around the vertical scaling topic and how hard it is. Seems your anecdotal evidence presented above backs this up.
I agree that we need to invest much more work in this area but also acknowledge the fact that the folks (Googlers) who drive this area (VPAs) in Kubernetes can bring a lot to the table, having experience running billions of containers per week over more than 15 years.
If you want to provide input and influence the decisions, I suggest you actively participate in the SIG Autoscaling, in case you’re doing not yet, already.