[Members] Messaging Interoperability initiative and the Digital Markets Act

Peter Saint-Andre stpeter at stpeter.im
Wed Jul 27 15:58:00 UTC 2022


On 7/26/22 5:57 AM, Matthew Wild wrote:
> Hi folks,
> 
> Yesterday I remotely attended an IETF meeting about forming a possible
> new working group focused on interoperability between messaging
> providers. 

Thanks for attending that - I hadn't been aware it was in the works.

> Here follows a little report on that, and a little insight
> into the potentially big changes coming for internet messaging
> interoperability in the coming years.

Sigh, it's the same problem we've been trying to fix since the late 1990s.

> In case you missed it, the EU's Digital Markets Act (DMA) compels
> so-called "gatekeeper" providers to open their doors to third-party
> access, both for person-to-person messaging but also other
> integrations (for example, maybe an application that allows you to
> analyze or export your data).
> 
> A "gatekeeper" is defined by a list of criteria, but the policy makers
> have crafted the criteria such that the definition applies to the
> usual suspects in big tech, without actually naming them directly. So
> this ultimately means Facebook, Apple, Google, and some others you
> might not immediately think of (Sony was mentioned, for example). Any
> small-medium sized independent XMPP operator certainly does not
> qualify as a "gatekeeper" under the DMA's criteria.
> 
> In theory the DMA is great. As many of us know, Google especially, but
> also others used to provide interfaces to their platforms using XMPP,
> but gradually turned them off and restricted access to their
> proprietary apps. The DMA is aiming to reverse this process, and
> require them to open up again so that third-party services and
> software can interact with their platforms
> 
> Rightly or wrongly, the DMA does not require them to use any specific
> technology or standard to do this (it can be argued that doing so
> would limit innovation). It is possible that some providers may choose
> XMPP, while Matrix folk are pushing for Matrix to be used, but by far
> the most likely outcome right now is that every provider publishes
> their own individual custom API.

It's not exactly clear how that will enable interoperability.

> This has many people (rightly) worried that a lot of the potential
> benefits of these providers opening up may be lost if nobody agrees on
> a common standard. Every app, service and platform would have to
> develop integrations specifically with every other app, service and
> platform, rather than just implementing one API/protocol and
> automatically gaining compatibility with every messaging provider.
> 
> In light of this potential future, the idea for a new working group at
> the IETF is under discussion. This group would work to identify common
> patterns across modern communication providers, and aim to, to some
> extent, describe reusable interoperable building blocks that the
> gatekeepers can utilize when opening up their platforms.
> 
> It is early days, and still at the stage of "what do we want to do"
> and has not yet reached "how do we do it". There is also the reality
> that even if we work on this, there is no guarantee that even one
> gatekeeper would adopt it.
> 
> Interoperability also has numerous technical challenges, still. The
> DMA requires that any API preserves common features like end-to-end
> encryption. Traditional gateways and bridges between protocols have
> always been free to translate between different wire formats, but with
> communication data being encrypted end-to-end, this becomes
> essentially impossible. 

MLS might help?

> Issues such as spam prevention were also
> raised as things that have come up as hurdles for large providers in
> the past.

Indeed.

> If all this is the kind of thing you're interested in, and you want to
> help potentially shape the future of interoperable messaging on the
> internet, the authors have requested feedback on the initial problem
> statement: https://www.ietf.org/id/draft-mahy-mimi-problem-outline-00.html
> 
> The mailing list can be joined at https://www.ietf.org/mailman/listinfo/Mimi
> 
> Either way, I hope this small report was useful and interesting to people.

Both useful and interesting. Although I've pretty much dropped out of 
IETF work (aside from a few "bis" RFCs I'm finishing up), I'd be happy 
to help advise you and anyone else on this initiative.

Peter



More information about the Members mailing list