Payloads for Twitter, round two
Sunday, September 30, 2007 by Dave Winer.
On Friday evening I wrote a piece about integrating images, audio and perhaps other types with Twitter. There's been a bit of reaction, not too much, I think because most of the people who are adversarial about this kind of stuff either don't use Twitter, or because it's the weekend.
Most of the reaction was either puzzled or negative. An example of puzzlement. Isn't that what Pownce does? Yes, but... Two things: 1. Pownce is still invite-only and 2. Pownce doesn't have an API, so it's inherently not as interesting to me, as a developer, because I can't build things on it.
I like Twitter because it's open to anyone to use, without an invitation, and lots of people use it, people I care about, and it has a very nice API. Further, as I've gotten to know the people involved, I've learned that the API is of supreme importance to them. So our interests are in-line there. I see Twitter as a framework to build things on, a platform, like a big Christmas tree we can all hang ornaments on. I could build nicer ornaments with a few extra wires on the network that connects all the ornaments. In fact, I've already built two of them, and we use them all the time. But I couldn't ask too many people to use them because they're too ugly. What I've proposed is a way to make them pretty, to make them work the way people expect them to.
Now another form of pushback is, well why don't you just build your own framework, different from Twitter, that does what you want, and leave Twitter alone. To which I say, I can't do anything to Twitter, other than talk about it. Whether to build the interfaces or not is up to the people at Twitter. I can have an opinion, yet ultimately the decision, and responsibility is theirs. Now, why don't I clone it? Well that's something I'm just not going to do. I have relaxed lifestyle these days. I'm beyond the point where I feel the need to prove anything through my work. I like to play and try out new ideas, just for the pleasure of it. If I were 20 years younger, I probably would be approaching this differently, but I'm not 20 years younger.
No doubt there are people, lurking in the shadows, who would like to share some of Twitter's success. The idea is so good that we're just at the beginning of its adoption. Maybe there are as many as 50,000 people regularly using Twitter. I think in a few years there will be millions, using Twitter, or something very much like it.
But we're at a unique place in the evolution of this stuff, which in some ways is very good. Suppose there were 20 Twitter-like systems out there, and we wanted to add a feature to all of them. Forget it! Developers just don't like working with each other enough to overcome their competitive urge. But right now, with one player in the market, we could make 10 times the progress we'll be able to make when there are 2 or 3. And a million times the amount in a market with 20 Twitter-alikes.
Further, the richer the API is, and the more broadly supported it is, the greater the incentive for newcomers to be compatible with Twitter. I don't get the warm fuzzies from Pownce that they are willing to follow anyone's lead, even though they don't yet have an API. But if anyone out there is brewing another entrant, and reading this, please please be compatible with the Twitter API. Not just the spirit, but the letter. Make sure that all the tools built for Twitter run without modification on your system.
So these are just some of the additional thoughts. Evolution of APIs is an art, not a science. I've learned a lot about it before the Internet, and then in XML-based formats and protocols. We're at a sweet moment right now, and if the Twitter guys want to lead, and if the rest of us are willing to be led (I am) then we can really build something wonderful.