I had a conversation about real-time search recently, with a journalist, and here's the question that I was asked: Does real-time search solve a problem?
And my answer was: No, not really. I think that real-time search solves half a problem.
I think that the real-time web is primarily about discovery. The "reference web" is about stuff that I already know exists. In the reference web, I search because I already know it's out there, I just don't know all the details that I need to know.
This is true whether I'm searching for a historical place or event (which often ends at Wikipedia), or if I'm searching for a web site (so-called navigational search), or if I'm searching for details on some product that I'm thinking about buying. All of these searches involve me knowing something, and wanting to find out more about that something.
In traditional reference search, when I first arrive at the web page that contains the search box, I already know what I'm going to type into it.
With real-time, my experience largely starts with discovery. I don't know what's new, noteworthy, or popular, right now, out there on the web. So when I first engage with a real-time web site, I need to be told what's "out there" on the web. Whether the new information is about a celebrity, an earthquake, or a product recall, I first need a summary of what's going on -- and that summary informs me with respect to any particular search I might then do.
When I first arrive at the web page that contains a real-time search box, I typically have no idea what I'm going to type into it.
So that's why I said, in my conversation with this journalist, that real-time search solves half a problem. It's an important half, to be sure, but focusing just on real-time search leaves out the key part, the starting point, which is real-time discovery.
So let me use this observation to springboard into a couple of brief predictions about what will happen to the entire concept of real-time, in 2010.
First, the entire web is real-time. It's not just about search. There's real-time discovery, real-time recommendation, and real-time contextually appropriate push. All the old-school activities that used to happen on a timescale of weeks and months will (or already are) happening on a timescale of seconds and minutes. We need tools for engaging with this accelerated world of information, and of course, that's exactly what Wowd is doing.
Another thing that we'll see in 2010 is that searching the space of tweets will take on its own special character. It's a form of vertical search, like automotive search, where you want to be able to search on some specific structured attribute of a vehicle, like a VIN, or body type, or transmission type. Essentially, there's structure to the stuff you're searching through, and systems that exploit that structure can deliver a better user experience.
In Twitter, there's obvious structure that can be used in search. There are people who issue tweets, there's the time of a tweet, there's possibly the location from which the tweet was issued, and there are people who are following the twitterer. And so on.
Searching effectively through this sea of information has special requirements, where a user might want to search for something by specifying some combination of "tweet must contain this text", and "twitterer must be followed by this person" or "tweet must have been issued in this time frame". And so on.
2010 will be an interesting year for real-time. We'll see increasing diversity in the ways that the real-time web can be embraced, and we'll see increasing power and flexibility to the way that Twitter-specific search can be performed. It'll be an interesting year. See you there.
