January 16th, 2008

Centeralization vs. Decentralization: What’s The Value Of Twitter?


Although I haven’t had time to use Twitter recently, I’ve been following all the discussions about Twitter, because it’s become a flashpoint for innovative thinking about online communication and media.

Twitter was much maligned for going down during Steve Jobs’ Macworld keynote — I agree with Larry Dignan that this is unfair. If you’re going to use an early-stage web service created by a start-up, then you have to tolerate the hiccups — but that downside is well worth the upside of riding a wave of (free) innovation.

But what really got me thinking was Dave Winer’s post about a decentralized Twitter:

Twitter is doing us a service, with its lack of stability, in illustrating the dangers of centralized systems. We do need to figure out how to build a Twitter-like system with all the advantages of centralization and none of the disadvantages. (Original idea from Andrew Baron.)

Dave suggests using RSS, with the right software interface. The advantage would be everyone could generate their own Twitter feed, which each user could aggregate in their own RSS feed reader.

Of course, there’s already a piece of software that lets you post content to an RSS feed, including via mobile devices and even IM — it’s called a blog.

If we all created a separate feed on our blogs for really short posts, would we need Twitter?

Fred Wilson has often sung the praises of Twitter’s lightweight nature, but I doubt even he’d be willing to invest in Dave’s decentralized Twitter. It would be like investing in blogging instead of in WordPress.com.

For all its lightweightness, what gives Twitter mass as a service, a company, an investment, and potentially a business, is its database. Which is also its liability, when that centralized database become inaccessible, but that’s true of all web services.

Without centralization you have, well…the web. And Google has already cornered the market for monetizing the open web.

Twitter’s software may be lightweight, but that also makes it a commodity — anyone could reproduce it — and that’s exactly what competitors like Pownce have done. Twitter has pioneered integration with SMS and IM, but that can also be reproduced — including with blog software.

So what makes Twitter Twitter? Its brand. Its people. And its database.

Twitter has come become the brand name for microblogging, which is one reason why so many people have chosen Twitter over Pownce or Jaiku. It’s a self-reinforcing system — just like Facebook and MySpace — the more people use a social service, the more other people want to be where there friends and colleagues are.

But it’s Twitter’s database that makes it a company and an asset — it weren’t for Twitter’s database, the Twiterverse would be like the Blogosphere — a loose affiliation that no one owns. I can understand why Dave Winer and many others would want that. But I can also understand why Twitter’s investors probably wouldn’t.

It’s Twitter’s database that stores all of the people you follow, the people who follow you, the people who you follow follow, etc. — i.e. the Twitter social graph.

Come to think of it, with all this beating up on Facebook over data portability, and the glorifying of Twitter’s API, where on Twitter can I, as regular (non-API) user, export the list of all my followers and followees?

Hmmm….could that be Twitter, the paragon of open platforms, locking in my data?

Of course, data lock in isn’t really the issue. Twitter could unlock all of its data, as could Facebook, and most users would stay because of the brand and the people.

The real issue is the centralized database.

Name one successful web company that doesn’t have as one of its core assets a big, honking database:


Yes, you could point out again that the web itself is one big de facto database. But, again, nobody owns the open web. And nobody has found a way to beat Google at monetizing the value of the open web.

And hasn’t one of the core principles of Web 2.0 (as a school of thought, not a school of hype) always been the power of combining data on the web the once lived locally on a desktop?

Umair Haque may be right they we are evolving rapidly towards a truly decentralized “edge” economy.

Then again, Jacque Derrida would tell you that you can’t have an edge without a center that defines the edge. (Would it surprise you to learn I was into deconstruction and critical theory as an English major?)

So for now, I’d keep betting on the centralized database.

Comments (13 Responses so far)

  1. What makes Twitter uniquely attractive is the multiple ways to send and receive “twits.” You can send receive twits via your cellphone, IM, email, and through a web interface. This means you can be plugged-in to the twitter network 24/7 whether your at your computer or not.

  2. [...] Karp rightly says that the people that have invested in Twitter will not be too keen on decentralisation. The major asset Twitter, or any other web service, has is their centralised user database. And [...]

  3. [...] if it were decentralised, but there’s a very interesting post by Scott Carp arguing that the real value of Twitter is it’s centralised database that stores the social graph that makes Twitter so [...]

  4. it all seems like such a dream from here in south india, a usb modem with a telcom that downloads at about 6 kilobits per, uploads at about 3, on a good day, i haven’t a clue what my cellphone would be useful for on twitter, anyway, the sms thing is usually an interuption

    its like you guys adopt, analyze, and move on, with technologies the rest of the world has no idea about

    out in the cold

  5. If the sum total of ideas that is Web 2.0 is “corporations are in charge of your data” then it’s hardly “2.0” – it’s “Client/Server Computing 1.0″.

    A truly Web 2.0 Twitter would be interoperable with a variety of other services, including locally-based versions. That interoperability would see “Twitter” (the service) as one of many services, all capable of talking to each other, based around a net topology rather than the kind of spoke-and-hub system it currently uses.

    In other words, truly Web 2.0 – or is this Web 3.0? – services would never totally go down just because a single server dies, just as the net itself doesn’t go down if a single router dies.

    “And nobody has found a way to beat Google at monetizing the value of the open web.”

    And why does every service need monetizing? The only reason that you need lots of money to run a service is if you base it on having a single, central database owned by you – because then, you have to pay for that database, and it’s bandwidth, and so on. There are many other ways to distribute information in a way that’s decentralised and that reduces costs to a minimum – like P2P.

  6. Scott: Twitter contacts are marked up in hCard, which allows services like Dopplr to do a simple read of the public page and make matches on their system. While not a true consumer facing export, it is a step simpler than an API call and is in keeping with what makes Twitter useful (and valuable): it’s open to extension, remixing & reuse with other pieces of the Web.

  7. Oh, and with a browser extension like Operator you could actually harvest the hCards from any Twitter page. Not just your own.

  8. [...] Twitter RSS feeds individuali ed aggregabili. E questa proposta viene commentata, tra gli altri, da Scott Karp su Publishing 2.0, il quale sottolinea che l’attuale forza di Twitter si trova proprio nel database [...]

  9. Hi Scott, I see Twitter being used for professional reasons over here. The ‘web experts’ are connected together and tell + read about who is doing, what. And helping each other out by sending links or asking + answering questions.

  10. Reading this reminded me of Scott Karp’s piece as well…particulary about harnessing network effects. What makes users stay with Twitter? The people (read network effects).

  11. There is a buzz in back corridors of IT industry that all these dot com companies do is disintermediation to re-intermediation…

  12. If Twitter had better redundancy and contingency planning, such that it had a back-up plan or system in place during the Jobs keynote, would this conversation even be occurring?

    Decentralization is one thing. Having a back-up plan to keep you going through a critical failure is another. They are separate issues that, IMHO, shouldn’t be confused.

    Twitter’s lightweight interface, companion apps, and ability to tie multiple feeds together fairly seamlessly are strong selling points. However, it’s the community and the appearance of greater and easier access to people who you couldn’t normally talk to (in real time, that is) which are the real value.

    On the surface, Orkut seems a lot like Facebook, without 99.5% of your friends, colleagues, or any other users, for that matter. (I’m generalizing, but I don’t think I’m far off the mark) That’s what differentiates Facebook from other like services. A massive community of communities built up around it. Twitter caught the attention of existing communities and hit/is hitting critical mass before the others, even though they have some superior functionality.

  13. I think Kyle nailed it with the first comment. Twitter’s innovation was not micro blogging per se but rather the multiple options for connection allowing for continuous interaction on the run. Youtube was not the first video web service. The portability of the videos posted to Youtube was what made the site go.

Add Your Comment


Receive new posts by email