EC: 'Guidance to Google for AI Interoperability on Android'

I very much understand where the EU is coming from with the DMA. I think that its intent, broadly, is a good one, and something we should be pushing for more. I'm even very sympathetic to the idea that platform-makers/gatekeepers should not just be able to charge ahead doing whatever they think is OK without any countervailing forces.

However, I think their implementation has, at least in some high-profile instances, been utterly boneheaded.

The problem with all too many such things at a governmental level is twofold (well, many-more-fold, but two major issues):

First, they treat "competition" as an axiomatic good regardless of the welfare of consumers. This can be seen in ignoring privacy concerns inherent in giving arbitrary developers access to the data required for this level of interoperability. I'm generally an Apple proponent, and I laud them for their privacy stance, but frankly I'd be more willing to trust Google alone with my data than Google plus Meta and whatever other developers want to farm it on Android.

Second, they barely seem to understand the technical details of the systems they are trying to regulate. As Gruber notes, they seem to be essentially putting themselves in the role of OS designers here, without the expertise to actually design an OS.

Finally, with regard to the DMA specifically, it seems really, really stupid that it's not set up such that an existing or potential gatekeeper can come to them a question about whether XYZ new feature would be OK, and what would be required to make it OK. As I'm sure many of us here know, it's vastly easier to build in interoperability and other similar features when first designing a system than bolting it on after the fact.