This message was deleted.

It’s very… opinionated. I don’t think some of what he says is wrong but it’s highly dependent on the organization you work at, the way the devs operate, the security around the application, etc, etc, etc. I don’t think having a monolithic platform that every “X” you build is required to use — but there needs to be a mechanism to put in place those guardrails he’s talking about at scale.

Horses for courses. My attention locked on “With some very rare exceptions (huge 1B+ focused companies) they just don’t work” — the question is really one of scale. The smaller the company, the more you should outsource by using higher-level commercial abstractions. The bigger you are, the more opportunity you have to find profitable economies of scale by doing it yourself further down the stack. (FWIW, I think the poster is miscalibrated about how many companies are big enough to justify an internal platform team.)

Well, that is certainly a provocative statement (probably hoping that it will end up like one of Kelsey's highly quoted tweets!).

Someone could build a platform out of this.

Actually, isn't this entirely the point of us building Minimum Viable Platforms? Once our platform features stop adding value to Dev (or they can build them better and more quickly than the platform) then we are doing too much.

Clickbait strawman argument that depends on defining platform in a way that tries to hide all the abstractions. There is no reason you have to do that. You can build a perfectly good platform that combines good defaults but allows you do override parts of it on your own. If an IDP means you need to send a ticket to an unresponsive group who won't act on it for 2 quarters, thats an organization problem and a design problem, not an IDP problem.

I shared my opinion here: https://twitter.com/manupaisable/status/1534867789323898881?t=gMYOpPWy3CK7C4cMefBM7Q&s=19