As I write more about my own projects (past and present), you’ll see that I generally root for the startup that can fight its way into a market and build value. There are times when, however, a startup should realize the larger role it plays in an ecosystem and set some of itself free. Twitter is such a company. In my previous post, I suggested that the Twitter service really should be cloned and released under an open source license. In this post I want to expand on the why and how…
It’s ICQ all over again
Making a communication service profitable is not for the faint of heart – direct response advertising in such a social medium performs poorly, and charging for the service generally only works when you can be cheaper than the alternatives. Most importantly, a communication service won’t have a chance surviving unless there are just gobs of users. So why a new service comes along and tries to build a closed system, preventing it from attracting users in every way possible, is beyond me.
Twitter is the ICQ of this decade, folks: a great communications idea that doesn’t make money and doesn’t work very often. And both Twitter and ICQ have chosen to be closed. When ICQ gained enough attention, it begat competitors like Yahoo! and MSN to make their own (closed) IM systems. “Hey I’m on AIM! Oh noes, I’m on MSN! Guess I have to make an AIM account and run both!” Retarded, yes? So how is this different from “follow me on twitter – no wait screw twitter, follow me on plurk!”
So I want to expand my previous plea of making parts of twitter open source: not only is it the best way to build a scalable service, but it allows twitter and its competitors to focus on how well it designs a service for its users. There is nothing sad or unfair to twitter about this; they are toast if they don’t do it.
The Conversation has started
First off, go read Eran Hammer-Lahav’s great set of posts on the topic. I agree with both Eran and Scott Hanselman that the computational challenges may be alleviated by focusing on smarter clients and a more sophisticated API. If I blog and produce a piece of XML, and your reader pulls that XML on demand, then the information is simply duplicated. Is there a role for a centralized server? Well, you could build an email gateway (you email your blog, and it can email you updates of what you follow), an SMS gateway (that you may have to charge for), gateways to other protocols, and of course you can build a web-store of all this data. Maybe these are the things Twitter Inc. should focus on (still not sure it’s a business though).
While we are separating the science from the service, let’s ensure the marketplace is interoperable. So why can’t I follow bob@twitter while posting to dan@pownce? Maybe a URL is all that is needed to identify the user from the service.
Sometimes you have to open up to your hard problems
Lots of people are talking about this — I suggest you follow how the conversations progress. My curiosity lies in what this has to do with open source and business.
Two proven capabilities of making any project open source is (1) if it is a worthwhile endeavor, you get really smart peer reviewers to help solve hard problems, and (2) it takes care of the not-profitable-but-necessary things like interoperability so that business can focus on how they bring value to their users. Open source has let products ranging from TiVos to iPhones focus on what consumers care about (which isn’t file systems and schedulers). Competing on uptime is no way to attract a mass audience, so why not let the fundamentals be a known problem and allow businesses to focus on building unique value? Let’s separate the service from the science. As popular as Twitter is, it still is far from mainstream. Twitter needs to cross the chasm, get people to understand the service, use it in as many ways as possible, build/support all the periphery services to it, sign carrier promo deals if they need to. All of this is really hard work, and it is assumes the underlying mechanisms are in place and ready for the scale. Once you have Skype-like numbers, then you can do some interesting brand advertising deals (”follow the new chevy on twitter!”).
A call to action
So this post is not much more than a call to action. I don’t know how to build the science and I am not interested in starting such a service — that’s up to all of you. But good business is always about learning from past mistakes, so let’s not let 10 years in Internet time obscure those lessons.
P.S.: I gotta plug Twoorl: Open source, nicely scalable, and written in a single day by an Erlang stud.
(Cat nap courtesy of a napping twitter image)