Twitter-like micromessaging is a relatively new communications model, with unique characteristics that affect how we use it and what’s appropriate. It’s an RSS feed for people, a way to directing the attention of audiences, and a means of reaching the famous without burdening them with an obligation to respond.

In short, Twitter is a human API. It’s being defined in real time in front of our eyes, through an amazing example of Internet Darwinism.

I’ve been spending a bunch of time on Twitter lately, partly because it’s fun, and partly because of the community management and social networking portions of a book I’m writing with @seanpower. Here are some observations so far.

Asymmetric follow

Twitter is neither one-to-one (unicast) nor one-to-many (broadcast.) Call it sometimescast.

This produces strange behaviors. The highly followed selectively amplify messages, and the less followed send nuggets of wisdom to the more followed in the hopes of getting their attention. This leads to the kinds of power laws we see in blogging. We probably won’t properly understand it until we have a pagerank for Twitter.

It also means there’s less context for a conversation, since we don’t easily see who’s talking to whom. So while back-and-forth chats happen, they’re usually brief. As a result, Twitter messages tend to remain a mix of instant messaging, URL referrals, bumper-sticker humor, and status updates. The lack of conversational context stops Twitter from devolving into a partyline.

Extensible syntax

Twitter didn’t have much functionality built in (the D convention for direct messaging was pretty much all it offered.) This meant the community created and selected its own conventions: Retweeting (prefixing a message with RT), @naming, and #hashtags started out independent of Twitter, and only later became part of the system.

HTTP was the same way. We had a few very simple verbs (GET and POST, mostly) and a lot of possibility. We came up with cookies, embedded content, URI parameters, and so on. They were optional. This allowed browsers and web servers to evolve commensally rather than requiring them to be in lock-step, resulting in much faster adoption and innovation.

Fluid relationships

Twitter’s explosive growth is due in part to the fact that anyone can follow anyone. It’s just as easy to unfollow someone, and we’ll soon see sites that track who’s being followed or unfollowed the most (think of Technorati’s Interesting Newcomers.) This keeps people honest; there’s always a lingering fear of being unfollowed.

The other day, I posted my first tweet with a swearword in it. It was a surprisingly emotional moment. I had to consider why I was swearing, and what it would do to my social graph, and whether I cared. In the end, I decided I didn’t care — the people who know me can follow, and the rest of them, well, they can leave.

This notion of whether you care if you’re followed is an elephant in the Twitter room. People want to seem smart. They want the affirmation of retweeting. They want to be noticed. Like it or not, the fluid social graph brings about yearbook psychology way down in our high school psyches, and has more of an impact on our behavior than we think.

The constraints of brevity

Developing for a constraint like 140 characters has an important side effect: Short messages make traditional brand marketing hard: Spamming a slogan on Twitter is a useless message, and a bad idea gets you unfollowed fast.

This has kept the medium relatively spam-free despite the fact that it’s so open. Other, richer social networks have to content with far more spam; but it’s hard to hide a virus in 140 characters.

It’s also responsible for making people succinct, causing them to prune their thoughts, which allows their audience to process more ideas with less effort.

An open API

Twitter has made it very easy to extend its functionality, in part because it hasn’t focused on monetizing the system. This means there are quite literally new Twitter apps going live every day (and the top ten lists to prove it.) There are also dozens of desktop clients.

Some of these sites are hashtag trackers, some tag clouds, some ranking tools. The point is that there’s a development ecosystem surrounding Twitter that’s unprecedented, and it’s creating new ways to extract meaning from the thronged masses. So even if individual conversations won’t get the attention of a company, a groundswell of objections will, right, Johnson & Johnson?

Many ways to participate

We want to talk to our friends face to face; we want to gossip about our enemies in public.

Hashtags allow transient groups to form around a topic. So if I hate a particular company, I want to get together with others and bitch about them to feel reaffirmed. But if I love a company, I want to personally connect with them in the hopes of their affirmation and acknowledgement.

Did I miss some? What other fundamental patterns are driving Twitter’s growth and innovation?