Has everybody already forgotten that OStatus¹ is a federated, open, production-ready protocol for Twitter-like notifications? Seems like building an OStatus-based clearing-house for messages and charging based on message volume would be a more sensible thing than trying to start a whole new ecosystem.
Dalton says a lot, but there's the sneaking suspicion that he wants to control the space. Personally, I'd be happier with:
- A charitable organisation handling the registration of names for multiple twitter like organisations. (Like DNS)
- Multiple providers (free, freemium, paid)
- Some way of routing messages to the correct provider.
- Some way of transferring messages between providers.
We managed this 25+ years ago with email. There's no reason it couldn't be done again.
As thristian posted, this has already been done, and it's called OStatus.
And you don't need a charitable organization running something like DNS; you can use DNS itself. The user ids just become like email, e.g. username@server.tld.
The original post by Dalton mentions freemium services such as Github and Dropbox as examples, so following a similar model and charging for certain features might allow for network effects to happen (although those examples aren't nearly as dependent on that).
I can see a usage based charge system where a certain amount is free, and then you are charged based on usage after that. Imagine a Heroku model for Twitter. Many users can get by with the free model, while businesses will likely be the ones going over the free limit and into the pay-per-tweet area.
Imagine you get 3,000 posts per month free. That's an average of 4 posts per hour every 24 hour period (6-8 more likely, because sleep). Most users won't come close to that, but businesses likely will depending on their usage and nature. You could even go the wireless phone route where you buy a subscription of messages in blocks, like 6,000 or 10,000 etc. This allows for monetization without punishing those who are experimenting or casually using the service.
I like this model, because then you have a scarce resource (albeit artificially so) that costs you nothing to make more of and you can use it to encourage users to do stuff you want, like inviting others. You could very well do it Dropbox style: you get 500 posts for every friend who signs up, and they get 500 too (starting out with 3500 in your example). It's not spamming if there's a gift.
If there's a "Like" or "retweet" equivalent you might also get extra posts for that (like you get more credits to post questions on Quora when people upvote your answers or follow your questions).
Surely, this would just lead to App.net being hated like the mobile carriers?
Price-per-x forces users to think about how much x they use. This is only a good idea if your users (a) need x or (b) are already hooked on x (say, by a couple of years of free x).
¹ http://ostatus.org/sites/default/files/ostatus-1.0-draft-2-s...