> A protocol by itself indeed doesn't need a business model but clients for the protocol do need one. They still don't need one, there can be many. Even if mos

> A protocol by itself indeed doesn't need a business model but clients for the protocol do need one. They still don't need one, there can be many. Even if most people still end up with the gmail model, it's meaningfully different, because the switching cost is lower. You don't have to convince everybody you know to switch too. Now suppose you're trying to solve the original problem. You've got feeds from a hundred users and you want a way to filter them yourself. If any service exists that provides this, using any business model, you can use it even if hardly anybody else does. You can also develop your own client that does this or commission someone to add the feature to an existing client. More than that, if this is a feature many people want, now the ad-supported providers will need it too, because now they have to compete for users with the providers who are treating you as the customer instead of being insulated from competition by the network effect when they own the entire network.