Thoughts on Real-time Search

Several companies have entered the arena of what has become to be known as Real-time Search. I've had a bit of problem really getting my head around what real-time search is and how it would weave its way into my behavior patterns. For me, Search is just one part of how I find out about stuff:

1. DISCOVERING: Something is happening, it may or may not be something I care about, and I don't know it's happening. I usually find out about it via some source, personal network, facebook, twitter, dig, CNN, etc. I don't have a specific mission to know it before I find out about it.
2. ALERTING: Something is happening and I knew ahead of time I wanted to know about it when it happens. I usually find out about these things via some source such as Google Alerts or Filtrbox. Alerts behave a little like a broad based search with asynchronous results that come back some day.
3. SEARCHING: Something is happening or has happened or simply exists. I want to know more about it. I generally want the best answer, or the most recent answer. I may want the best recent answer, but that's highly subjective and generally defaults back to my trust in the source as the tie breaker.

For the record, I think Google does a pretty good job of real-time "search". You have to know how to apply date bias to the search but they index very quickly and have larger reach than anyone else. Of course, I have to know that I wanted to know which means the actual SEARCHING has to be paired with ALERTING or DISCOVERING.

It appears that most of the specific Real-time search services seem to combine SEARCHING with DISCOVERY. For me this moves the combined activity into an entertainment arena, rather than a fact finding arena. It's like buying a magazine about Coffee, now Milk Dudes, now Bottled Water – interesting but not entertaining.

It seems a lot of real-time search services are paired with Twitter. For me this makes a little more sense, first because twitter is largely entertainment (at least to me). I'm really not there to learn anything except what the people I'm following are doing. Twitter is however heavy in ALERTING as was witnessed in my last post.

Twitter (ALERTING) and SEARCHING seems like a more natural combination than DISCOVERY and SEARCHING. For instance, when I get a Tweet about something, it would be cool tap the tweet in my iphone and return the most recent 5 items from the web that match that Tweet. For instance, "Michael Jackson dead", would return the top results with date bias from Google or another search engine.

Another use case that makes sense to me, "Michael Jackson dead" returns the latest news from the online sources that just people I follow read, sorted by date. In other words, I got the alert from my network – I want the most recent news from my networks sources. This is the Lijit spin on how to find information.

My takeaway is that there are combinations of the three activities DISCOVERY, ALERTING and SEARCHING that I can make sense of in certain situations. I just can't get my head around the combinations being a "task" that I perform on a destination web site.

It only makes sense when they are combined organically in another place where DISCOVERING or ALERTING is happening.

Comments

[jvaleski]

I'd dumped the following into an email a few weeks ago. Thought I'd paste it here as we're seeing some similar things/use-cases/patterns.

"real-time search sucks from an end-user product offering standpoint (every service to-date is effectively useless though (disclosure I'm a shareholder in oneriot so want them to succeed obivously)). other than a somewhat interesting view of "trending topics" (twitter, oneriot, collecta, etc), the only end-user use cases that can be remotely solved right now are: one, "show me spike events that ppl are talking about right now" (e.g. earthquake), and two: "show me the topic cloud of what's going on right now"). the former use case is useful when you hear that something just happened ("plane crash!"), the latter use case is a snoozer. do you find yourself wondering "gee, what's trending right now"; probably not. let me put it another way, do you ever use oneriot or collecta for searching information? the initial value in the model is in "what's hot right now."

until there's purity/consistency in the streams of data that are being searched, this is going to remain a problem for RT search companies; highly limited usefulness."

[todd vernon]

Jud, great comments. I think a lot of what you point out comes back to "this is entertainment". There's nothing wrong with that of course, but its not the intent stream that has been harnessed by someone like Google and will likely not monetize as such.

Since I wrote this post Google has announced their social+real-time deal with twitter. I think this is a killer deal (mostly for Google, maybe for twitter) as it's the worlds most powerful social graph connected to the worlds most powerful search engine.

That said, Google probably won't mash the gas on it, BUT I think this certainly steps on the air-hose of others in the 'destination search' game that want to connect those dots. IMHO.

Google IS great destination horizontal search and Twitter is great "meaningful" social graph. There you go.