I see this as more of an opportunity than a problem. The fact that Hadoop, Kubernetes, and other platform-like systems are complex to manage properly with good attention to security implies they should be delivered as cloud services rather than having everyone run their own. This enables K8s users to focus on apps while offloading management to specialists who can focus on running the services well.
If you operating at large enough scale you can bring the "cloud service" in-house.
One of the gating factors here is the both the speed of Kubernetes development (move fast and break all the things), and the terrible state of accompanying documentation.
If "they should be delivered as cloud services" is some sort of k8s apologist stance for its sorry state of maturity, then we got issues. OTOH, if it's "You shouldn't run it in house unless you have an army of people to read every new commit", that's wrong, too.
First of all let me be clear I'm not an expert in K8s and certainly not an apologist for bad software.
On the other hand there's a level of complexity in distributed systems that is impossible to avoid even in stable infrastructure. You have a design choice of trying to make the system as easy as possible to operate (at the cost of other features) vs. finding operating models that make it less of an issue.
Personally I would rather spend time futzing around with my applications that run on kubernetes vs. trying to run kubernetes itself. It would be sufficient if Kubernetes services were portable across a marketplace of providers so I could pick a place to run my applications.
Also as far as security is concerned it appears to me that a lot of people are deploying technology that they simply don't understand. This is not just a problem with Docker but with apps from ecosystems running on npm and pip.
You can build images securely with Docker but it requires building them yourself, using private registries, checking carefully for vulnerabilities, and testing. If you don't want to do this, pay somebody else to do it right. There's no free lunch.
If you operating at large enough scale you can bring the "cloud service" in-house.