This message was deleted.

Nice piece. The part that tripped me up was the bulleted section that declares that "all platforms are..." Myself, I like to think about it in terms of the CNCF platform maturity model, how different aspects of the platform may be a different levels of maturity. That may actually be a feature of any org, but at least it gives a framework for thinking about where you might want to invest (or possible divest) next. https://tag-app-delivery.cncf.io/whitepapers/platform-eng-maturity-model/

I included the four restrictions because it's important to distinguish platform engineering from what has come before. There's a strong push to expand the definition of "platform engineering" into something that's big enough to cover basically any kind of automation or CI/CD. As I said in the piece, platform engineering IS something new, and we need to talk more about why. Automation is not new. We're talking about more than that.

I'm not sure I communicated my point as well as I could have. I understood your four restrictions to be qualities that if they were absent would mean that something was not a platform. In my view, a platform is a platform even before it has been "productized." (I also suspect that my definition of "productized" is different, as I see both user-centrism and self-service as qualities of a healthy product.) IMO, paved paths can enable their desired outcomes even if the solution isn't mature enough to have robust guardrails that insure compliance and consistency (again, comes with maturity). Is that out of alignment with your view?