Friday, July 18, 2008

Twitter's business model emerges?

For some time I have been thinking about Twitter's business model. The best clue came months ago when Twitter started playing with the throttling value for the API.

Since then it has become clear that a few select partners have access to the unfiltered XMPP feed. Summize, who Twitter recently acquired, being one of them.

Twitter has been struggling to avoid the fail whale for months now as growth surges. Louis Gray covered the latest change in his blog as Twitter chokes unauthenticated API requests by IP.

When Twitter first started playing with throttling my first thought was to their business model. I can see Twitter working like real-time stock quotes. If I want to get an "up to the second" feed then that is a service to pay for. If I am happy to wait 15 minutes then I can have it for free.

I think Twitter could implement this model relatively easily. Replies and Direct messages would be delivered to you immediately but your public timeline would be delayed 15 minutes, unless you paid for it. This approach might also enable performance tuning and scaling to be implemented. Imagine Twitter analyzing the feeds of the most popular users. It might be able to detect tweets that might trigger a tweet storm. With a 15 minute window they might have time to ramp up additional temporary capacity to manage the storm.

For sites that want to build their own business on top of twitter it seems reasonable that they would want to pay for a real-time feed - with service levels commitments wrapped around their feed. Twitter could even allow a capped real-time feed so that developers could test their applications. The cap could be based upon a set number of calls or volume of data. If you read the small print of Yahoo and Google's API services you often find a cap on the number of calls that can be made in a given time period.

So, Twitter, what are your next steps?

You seem to have bungled your handling of the throttling of unauthenticated APIs. You need to lay out a roadmap quickly. If you simply cut off the data supply to developers without giving them options then you will force them to define their own roadmap which will no doubt exclude Twitter and you will see an exodus to emerging services like plurk and This is the last thing you want if Twitter is to become another essential communications utility. You need to provide the developer community with options so that new services can continue to evolve in the Twitter ecosystem. It is the ecosystem that makes Twitter so valuable.

Twitter Acquires Breathing Space

Given the reliability troubles at Twitter the purchase of Summize was a critical strategic move. With the loss of the Replies feature in Twitter on occasions I have often used Summize to do a search on @ekivemark. May be Twitter will modify the Home page to make that same call.

As the leading search service for Twitter I would have expected Summize to come under increasing pressure to integrate search for other services. As federated micro-blogging takes off Twitter would have been marginalized. The acquisition nips that in the bud.

Ecosystem diversity

Dave Winer has been advocating for a Federated Twitter for some time. The latest throttling move may alienate the developer community and promote the federated model if Twitter is not careful. I have been waiting for the API interface from A plug compatible API for the federated micro-blogging world would enable the Twitter ecosystem to provide diverse routing. This would also open up a host of new client and middleware opportunities. Middleware platforms could provide a useful service "de-duping" messages that are delivered via multiple routes.

Adding the Meta Layer to the Flow

Stowe Boyd talks frequently about the flow. However we need to manage the signal-to-noise ratio in the flow. As middleware platforms expand they can address this need for the micro-blogging world. They can provide a meta layer that could filter the flow to provide context. For example, surfacing just the useful gems of comments from a noisy self-publicist, or filtering the flow based upon a combination of interests, current location, time of day and other contextual parameters. This type of filtering could get around the need to follow and un-follow people.

In the short term I see a lot of opportunity for services like that enable posting to multiple sites. I am starting to experiment with Ping to see if I can post to Twitter, Plurk, and BrightKite. The proliferation of sites is making it harder to keep up. I find Twitter and Brightkite to be my preferred sites at the moment because they the web based applications from Hahlo and Brightkite are a joy to use.

What do you see as the future for Twitter and the Micro-blogging world?

No comments:

Post a Comment