Jul 25, 2008


Twitter is an unusual,and I think misunderstood beast. On a simplistic level it is a service that is useful on its face value. I tried Twitter for a while and found it to be an intriguing communication tool; its primary purpose. There are many reasons I feel that way that I won't go into here.

On a deeper level however, Twitter has a number of characteristics that make Twitter, the platform, incredibly intriguing and full of potential.

First, many many people love the service. This is important because the amount of micro-data being captured is large. This in turn is important because it means that the highly diversified topics of content that are generated individually have weight to them; the data captured is statistically significant. Filter the content through any number of parameters like geography, time, subject and you'll still get a lot of data on almost any topic. So it has a reasonably robust data set and one that is getting more robust over time.

Second, because the data input is limited to so few characters it's remarkably real time. Content comes in remarkably quickly after an event happens.

Third, the content has recently expanded from human entry to non-human entry. This one is very important. There are now electronic components that twitter when a bridge is going up or a plant needs to be watered. Don't be blinded by the narrow use of data generated by these two examples. Imagine what kind of event driven data could be uploaded into Twitter.

Fourth, the data is all centrally located and distributable (as long as people don't block the feed). In other words you have a massive database which can be parsed easily.

So it creates a content or data space that is hard to come across with other types of platforms. Lots of things have one but not the other. Blog posts are highly varied in their topics but they are not written during events, are only written by people, and are not centrally located. Papers are highly specific, written by people, decidedly not real-time, and not centrally located. Instant messaging is real-time but it's not distributed. And so on.

So take a step back for a second and consider the platform. It's a message platform. Think of other things that are messaging platforms? Bloomberg terminals are messaging platforms. Data is messaged through to the termainsl or excel spreadsheets. The NYSE is ultimately a messaging platform. One for sending stock orders to a central database. The telephone is a messaging platform. Web services are a messaging platform. APIs are a messaging platform. I could go on.

These are all real-time, include people and machine data, and are centrally located.  But the breadth of data is narrow (albeit very deep in some regards). Think about that for a second.  Bloomberg and Reuters are very valuable properties.  In some ways Bloomberg is a small version of what Twitter could be.

So if you put those four characteristics together you get an unusual data set. And it's the data set that I'm most interested in and why I find Twitter interesting. The only service that has really utilized this data that I'm aware of is Summize.com. Summize was a search engine that searched Twitter feeds. And it did this in real-time because it's centralized. Google search doesn't really do real-time. Twitter was smart enough to buy Summize out early. This is an incredibly savvy move. This is a data set that Google must own if they wish to be the center of the universe of search. And now it is too late bar a partnership. Google will be forced into that partnership.

Summize for me is more addictive that Twitter itself. When a piece of unusual stock information comes across or a news item of interest I'll usually check the content feed from Summize. You find really unique information that hasn't (yet) been disseminated widely.

When I saw Obama in a hotel lobby a few months back I checked Summize. I found out he was staying at the hotel, I found out the floor he was staying on, and I knew where he went next.  Real time.

That's remarkable. And yet Summize is really the most rudimentary of services. I can't help but think that lots of useful services that mine that data and perform analytics on it are going to start coming out. Entire businesses set up around Twitter data. I'm not smart enough to come up with them but I see the value in that data set and how unique it is. Someone is going to come up with some clever mashup.  I'm not sure if the services like Summize will be monetizable the way Google search is, but they could be very useful to the point that they can command some value.  And I still keep thinking of the data set as a Bloomberg data set but only bigger.  And therefore potentially more valuable.

So I hope Twitter keeps on keeping on.  But for me it's not the service but rather the data that should be what the discussion is about. People who focus on it being a service to talk about what someone had for breakfast are being myopic.

No comments: