I am possibly simplifying a bit too much, but: Build world-class products around their protocol themselves!

I mean, if that's it... Imagine WhatsApp or iMessage allowing their users to talk to Matrix, but not make it as easy to find people there as on their own services. People who might've been tempted to join Matrix before would probably no longer be, because they'll feel they can get all that with their existing service, while they'd still pressure Matrix users to join that service because the experience is subpar from their perspective.

True, network effects are hard to beat, but one better hope the network builds around the product one themselves built around their federated protocol.

It works both ways. With an open network, someone can build a better client than WA or iMessage and then there is social pressure to converge on the better one. The best app ends up winning rather than the one with the biggest walled garden. The consumers win; complacent gatekeepers lose.

Thanks for chiming in! But the problem I see here is that WA or iMessage (or whatever incumbent) needn't be fully open on a network. They can give themselves access to the open network, without fully opening up their own network to the rest of the network, and then by virtue of being the incumbent, thereby prevent better clients from taking off. (And that, of course, is the embrace & extend part. Once they're the dominant player again...)

Our countermeasure is that in Matrix you can’t build gatekeepers, as conversations replicate between the participating servers. You can’t talk to someone on another server without literally sharing ownership of that conversation with them. Now everyone could chose to congregate on a single server (e.g. matrix.org has about 35% of the users on Matrix right now, given it’s the default) but there is no advantage to doing so.

Meanwhile we’re building out P2P Matrix to avoid being dependent in servers altogether: arewep2pyet.com