Today Twitter cut off Tumblr’s access to the Twitter social graph, preventing people from using the list of people they follow on Twitter to find friends on Tumblr. This is yet another lurch toward stinginess on Twitter’s part. They obviously feel like people are hooked enough on their service that they’ll stick around despite the fact that they’re claiming your list of Twitter followers as their own property, and have decided to slowly starve third party developers who build Twitter clients.

One alternative to Twitter is App.net, which has certainly started out by saying the right things but will either never catch on or become yet another proprietary network that finds itself in conflict with its users. The just-announced Tent.io is a proposed protocol that would enable people to create their own Twitter equivalent in a federated fashion, in the style of other popular federated Internet communication systems like email, IRC, Usenet, or even HTTP. With Tent.io we’d have a large number of “Twitters” that pass messages between themselves so that there’s no one centralized service upon which everyone depends.

The tough challenge they face is figuring out how to move from a protocol with no available implementations yet to a service where all your friends are. This service will, at least initially, require you to set up your own instance of the software on a server somewhere. If the service catches on, services will arrive that enable users to simply sign up on the Web and start participating, just like they can get an email account through one of any number of free email providers. But the effort required initially will be greater.

The way to get started is to write a Tent.io implementation that also works as a Twitter client, so that you can publish to both your Tent.io server and to Twitter at the same time, and so that you can read messages from both Tent.io and from Twitter in the same interface. Then if things go unrealistically well, traffic gradually moves from Twitter’s service to the Tent.io network.

The second important step is making lots of friends. Is it possible to get Yammer to implement Tent.io so that companies can federate their Yammer networks with other Tent.io networks? The people behind the Identi.ca service were talking about the federated social Web back in 2010, but their service hasn’t really taken off. They have an open source Twitter clone called Status.net. The folks behind Tent.io should be working with them and learning from them.

The third step is moving beyond Twitter in some important way that compels people to switch over. “Like Twitter but open” is not a value proposition for mass adoption, or even for significant adoption. Compelling features will be required.

This effort reminds me a lot of XMPP, the open, federated instant messaging protocol that was created as an alternative to proprietary instant messaging networks. It has been successful as a protocol, but not necessarily as a means of liberating people from proprietary networks. Google Talk is built on XMPP and supports open federation, but almost nobody takes advantage of it. They just use Google Talk. Worse, none of the other major instant messaging networks have ever bothered to add interoperability with Google Talk via XMPP.

In the end, the best hope is that Tent.io gains enough traction to be adopted by would-be Twitter competitors, thereby insuring that people do have some control over the network they create on the service, have access to the full archive of posts that they wrote, and can generally be insured that the investment they make in the service won’t be devalued over time through business decisions made by the service provider. An alternative outcome where people run their own Tent.io servers in the same way that they run their own blog software seems less plausible to me.

Twitter, which can in many ways thank third party developers for the great success it has enjoyed, is imposing wide-ranging restrictions on third-party developers going forward. They’ve been moving in this direction for awhile, and the latest announcement makes these changes official. Some time ago that made it clear that third-party clients are an endangered species — not only are they discouraging development in that area, but they have also been rolling out new features regularly without corresponding API calls. Now we see that they are imposing much tighter restrictions on API usage in general, and also restricting how content from Twitter can be incorporated into other Web sites.

My biggest complaint with Twitter’s evolution is that it feels like a bit of a bait and switch. More than most other social sites, Twitter owes a huge amount of its success to the creativity of its users and to its developers. Twitter developers have reliably stepped up to meet the needs of the community when Twitter’s official clients have failed to do so. Many of the interesting features Twitter has added were created organically by the community. @-replies, retweets, and hashtags were all bottom-up innovations that made Twitter more useful or more entertaining. Treating Twitter like a game was also a community-created feature, promoted by applications like Favrd and later Favstar.fm.

Now Twitter looks a lot like the big star who forgot about all the little guys that helped it get to the top. Twitter’s early openness was one of the major factors that attracted people like me to it over the closed, overly structured Facebook. Now, though, it looks like Twitter envies Facebook’s walled garden and wants to emulate it. At its best, Twitter has been “of the Internet” in a way that most of its competitors are not. I hate to see them intentionally throwing that away.

I’m not quitting Twitter or anything — I love Twitter. That said, I don’t think that the people who run the company are really in touch with what its users love the most about it.

I’d encourage you to read Marco Arment’s post Interpreting some of Twitter’s API changes for a detailed critique. It really helped clarify my thinking on today’s announcement.