Question about "kubernetes-resource-planning" blog article

[wsierakowski]

I have a question about the following piece in this article: https://goteleport.com/blog/kubernetes-resource-planning/

Another advantage is uniformity for deployment. An application may consume 80% of resources which makes the leftover resources too small to be useful for others. Hence, administrators may require a minimum of 25% of memory and CPU to provide uniform scheduling.

Should the admins in this case require a maximum of 25% of memory and CPU instead of a minimum in order to provide uniform scheduling?