eran’s posterous

 
« Back to blog

Twitter will be the web's Remote Procedure Call (RPC)

Over the weekend, I whipped up something kinda cool. A twitter account called @EventsInSF. Using Dapper and some php I've created a service that aggregates local events happening in San Francisco and the bay area and streams them into this Twitter account. It's kinda cool, and it seems to be useful to at least 50 people so that's great. But I think it's a signal of something much bigger.
Up until now, everyone were highly fixated about the "man-to-many" nature of Twitter, making it the greatest platform for celebs, PR people and immersing in the collective meme bath. @EventsInSF is, I believe, one of the first examples I'm aware of when a machine has a Twitter "persona" and is directly communicating with people on Twitter. As such, Twitter is a great content distribution channel as it takes care of the scaling. From my perspective, the same amount of resources are required to run @EventsInSF whether it has 50 followers or 500,000. How about that for good scaling?

However, there's something even deeper. What if machines will start talking to other machines using Twitter as their RPC platform? What if someone will build a service that will "follow" my @EventsInSF account and map those events as part of a travel mashup? Even better, what if it will send out commands, not only simple content? This loosely coupled nature of the Twitter "follow" motif and its close to real-time nature, are just perfect for it becoming the back-bone of autonomous machines talking to eachother over the web.

There still remains the issue of "service discovery". One could easily envision Twitter accounts being used as "directory services", or in fact using Twitter's search as a way for a machine to discover existing services of a certain type, holding certain content.

So what is my only issue with this rosy picture? It's the fact that Twitter is a service built inside a single company, that has full control over how people are using it. Envision a world without e-mail. Now think today someone will come up with this cool idea of a one-to-many push idea and name it "Twiemail". How comfortable would you be living in a world where there's that company that is the sole proprietor of your main way of communicating with people, and can ban you out any minute, censor your messages or what not? Why are we still comfortable with it when it's "one-to-many pull" rather than push?

Comfortable or not, Twitter will become a main channel for communication not only among us humans, but also for the machines that will supersede us. In a sense, as we're getting nearer in bringing the semantic web vision into life with services like Dapper, the time is getting ripe for a popular "glue" service to become the bus of the semantic web. Twitter could be just that.

Loading mentions Retweet

Comments (7)

May 20, 2009
 said...
What you're looking for is identi.ca, i believe.

P.S. There are quite a lot of twitter bots in the wild, including one that I wrote that sends me a DM every time one of my servers is non-responsive.

May 20, 2009
eran said...
Yes, identi.ca does the work technologically, however as long as it will have zero adoption we'll remain in that same situation where you must use twitter.
The big challenge is to move people into a federated system.
Loved your twitter as nagios use case :)

Sent from my iPhone

May 22, 2009
jjmajava said...
You might be interested in what @stoweboyd is suggesting as a new microstructure for locations http://tinyurl.com/p33erg
May 22, 2009
 said...
Identi.ca provides a safety net: if Twitter does something that most (influential) users deem objectionable, then a concensus to shift will emerge — leaving Twitter masters with only hours to react. They might be tempted to boil the frog slowly, but the clustered structure of Twitter and the possibility to built bridges with Identi.ca make it easy for specific luddite tribes to shift when they don't like that company anymore, until all are gone. What made Twitter is its ability to launch storms on a water glass; that same ability make it easy for users to switch. While hardware and most software decision have become seamless (May the fail whale be merciful!) usage change still trigger bursty reactions —as we saw recently.
May 22, 2009
eran said...
As we jews say, from your mouth to god's ears. Somehow, I find it hard to believe @aplusk or @theellenshow will join the revolution :)

May 22, 2009
trappermark said...
Before your Brave New World of automated systems communicating with each other through Twitter could become a reality, there is another concern besides the whims of Twitter's owners: Twitter's infrastructure. Bertil already mentioned how we are at the mercy of the Fail Whale. But the troubles run deeper than that. Witness Louis Gray's article about how broken Twitter's much-vaunted search engine is: http://cli.gs/UqZTVT

On a lighter note, as "facebook user" noted above, there are indeed many automated service already crawling the Twitterverse. Just the other day I entered a word in a tweet of which I was unsure of its spelling, so after it I put (sp?). Within seconds I got a surprise tweet from a bot that gave me the correct spelling.

May 22, 2009
eran said...
Indeed, I was totally clueless about the swarming of bots on the twitterverse :) Obviously I like the fact that I can use Dapper as the 'bridge' between the 'real' web and Twitter (perhaps we'll do something more generic about that someday soon). The question is, how much time will it take before there will be more machine tweets running than human initiated ones. Care to bet?

Leave a comment...

 
To leave a comment on this posterous, please login by clicking one of the following.
Posterous-login     Connect     twitter