Real Time Search Is Personal

The other day, I promised in a comment thread that I’d write about what I see as real use cases for real-time search. As it happens, I’m experiencing one right now.

As my wife, daughter, and I were walking home from a playground, we noticed a large number of fire trucks congregating a block away from our house. A quick search on Twitter explained what was going on, particularly by pointing us to this post on Gothamist–which as of this writing seems to be the only reporting about this incident.

I think this example tells us a lot about the utility of real-time search. Most of us don’t need real-time search to tell us about the news in Haiti, since a critical mass of major news providers is covering the story around the clock. Where real-time search matters most is at the personal level–specifically, when our personal urgency to obtain information is higher than that of the general population. In such situations, we’re willing to accept less polished–and even risk less accurate–information, particularly if the alternative is to wait until if and when news providers cover the story. At least to some extent, urgency trumps authority.

Yes, there are other use cases for conversational media like Facebook and Twitter, such as sharing the experience of watching a live event, or simply chatting with friends and strangers about arbitrary topics. But I wouldn’t consider such use of these media to be search. Real-time search, in my view, is about helping users obtain the latest information available–in accordance with their personal needs. Twitter and Google served me well today, and I’m grateful that real-time search gave me real-time peace of mind.

By Daniel Tunkelang

High-Class Consultant.

19 replies on “Real Time Search Is Personal”

I’ve heard leaders at Google say that a feature has to be used at least 5% of the time in order for it to be considered worthy of development time and effort.

So my followup question is: Given that you’ve demonstrated utility, what is the frequency of that utility? How many times per year do you need to know things like why fire trucks are congregating near your house? Is it at least 18 days a year? Or at least 1.5 days a month?

Or is it less than that? More than that?


I’ve never heard that 5% claim–have you seen it in writing? It strikes me more as a way of saying that Google prefers to make investments that will have a major impact–and that’s not always easy to measure quantitatively.

Large numbers of fire trucks outside my house are, fortunately, a rare occurrence. But I did have a similar experience recently when there was an earthquake close to where my wife’s family lives in Humboldt County–though that did end up being covered by national news outlets pretty quickly.

Perhaps a better question to ask is how often I’d use real-time search for even the most mundane of tasks if I could rely on it, e.g., to decide whether the local Trader Joe’s is too crowded and I’m better off going there some other time. Perhaps I’m lazy, but I’d appreciate that sort of information access if it were convenient.


See #8, here:

I guess I misremembered it.. it wasn’t 5%. It was the higher threshold of 20%. At least 20% of people need to use a feature in order for it to be included. 5% is the lower threshold for an option to make it to the advanced search preferences.

It’s obvious that RTS is a feature that Google wants to appear on the front page, not buried in advanced preferences. So I’ll ask it again: Do you really see 20% of searchers using this? Has 20% of the U.S. population been through an earthquake that they wanted to get immediate information on, in the past year? Has 20% of the population had fire trucks a block from their house?

Don’t get me wrong.. I’m not dismissing what you’re saying. I’m asking for further clarification. You give a good example. But is it the rule (does it exceed Mayer’s 20% threshold?).. or the exception?

Fire trucks are a nice example, but they’re the exception, not the rule. They certainly don’t push searchers to that 20% threshold.


And Trader Joe’s is a nice example, too. But is it RTS? Going back to my Garmin (I used that example the other day, too :-), I’ve been able to get real time traffic information here in the Bay Area for years now, and the unit will automatically re-route me in real time, based on the sensor network data that is streaming in. And I use that information 5 of 7 days per week. So it’s very relevant to me. And Trader Joe’s crowd levels fits into that paradigm seamlessly.. extend the sensor network to parking lots and stores, and “route” me through my daily errands in the most efficient way possible.

But is that RTS? Is that what is meant by RTS? I always thought that RTS meant people both creating and consuming textual information in real time. Searching Tweets, for example. Explicit user queries. Explicitly-created information (real people authoring the information that is being searched), not automated sensors doing the tweeting.

What I’m trying to say here is that it’s not that I disbelieve your answer.. it’s that I’m not yet fully compelled or satisfied by it. There’s gotta be something more. Everyone is so excited by RTS. I still don’t see it.

Do any other Noisy Channel readers have any ideas?


This is a great post and I enjoyed reading the back and forth in the comments, too. I believe that Real-time search cannot be looked in isolation. Search as a whole needs to be improved and the future of search will be more real time. Have any of you seen TipTop ? That, for me more than any other product, shows a glimpse of the Future of Search by bringing in good doses of real-time, social and semantic search in the right proportions.


Jeremy, thanks for the link. I don’t set those thresholds, but I do notice that she constraints how many people use a feature, rather than how frequently they use it. Do I think that at least 20% of the US population finds themselves in situations like my fire truck scenario at least once a year? My intuition is yes.

But you raise a fair point about the boundaries of real-time search vs. sensors. I don’t feel that all of the components have to be explicit. On the query side, I think that the boundary between search and alerting is fuzzy. Indeed, I’ve seen people refer to alerts as standing or continuous queries. And on the authoring side, I see no reason why a user should care whether the information was created by people or machines.

Perhaps I’m over-generalizing the accepted definition of real-time search. I would be curious to hear how other readers here define the space.


Shyam, I checked TipTop out. Interesting, though I’m actually impressed with my own employer’s integration of Twitter results. Though I’d love to see more support for summarization and exploration.


To the point about the 5% / 20% threshold. What’s interesting to me is if Google sets such a threshold — and then identifies certain offerings as not meeting it — they have just identified an opportunity for niche players, right? Given Google’s scale, even a rounding error in their world is a pretty good business for a smaller, more agile player.


Cool, Daniel. I am glad I stumbled on to your blog yesterday. We are a very small company just starting out. What is out there in public at the moment is only tip of the iceberg. There is lot more coming in the next few weeks and months. Please visit often and share with your colleagues and friends. You can already check out our TipTop Shopping link to see how well we summarize reviews on Amazon. Lot more exploration perhaps of the kind you like is coming out soon, too. I’d love to have a chance to chat with you sometime to learn more about what else you’d like & to tell you more about TipTop.


I don’t set those thresholds, but I do notice that she constraints how many people use a feature, rather than how frequently they use it. Do I think that at least 20% of the US population finds themselves in situations like my fire truck scenario at least once a year? My intuition is yes.
Sure, I would agree that, at the 1-year granularity, you could get 20% of the users onboard.

But what Mayer doesn’t mention or constrain, at least in this PARC talk, is the time window over which 20% of the people will have used the feature. Is 1 year really the window? That seems awfully big. I used it as an example, just to question how many times a year someone might use the feature. But from what I’ve heard about Google A/B testing, most trials don’t last longer than 2-3 weeks, am I correct? Iterate rapidly, iterate often. Not iterate once a year.

So do at least 20% of the US users need a feature like this, every three weeks?

But look, I’m becoming guilty of slicing this discussion a little too-finely. Let me recant. Let’s go back to my original question: How many times per [choose whatever time window you want] does the typical user actually need/use this sort of search?

I do 5-6 regular web queries a day. Or 1500 per year. How many RTS queries would I (or some other average user) probably do per year? More than 1500? Same? Much less? My point is that even with the firetruck scenario, I don’t really see heavy usage.


On the query side, I think that the boundary between search and alerting is fuzzy. Indeed, I’ve seen people refer to alerts as standing or continuous queries.

Oh, I think I agree with you here. Active or passive queries.. doesn’t matter.

And on the authoring side, I see no reason why a user should care whether the information was created by people or machines.

This is where I’m not so sure. I see machine authoring as conceptually very different than human authoring. With machine authoring, I totally see a daily need and use case for RTS. Like I said, I’ve been using RTS for years now with my Garmin traffic re-routing software, which makes use of sensor network “tweets”.

But that’s been around for years, without spawning all this current excitement. No, the way folks are talking about RTS is because of Twitter. Human authoring.

And I still don’t see it.

But I agree.. what do other readers think?


Very interesting comments from you, Jeremy. I think lots of questions swirling in your mind, Jeremy, can be answered if you spend a good one hour on TipTop I am also happy to chat with you sometime so that we can discuss this further.



I played around with it a little bit, but it still doesn’t really address my issue. My issue is not whether you can get good answers (or even good exploratory information), if you have a real time information need. My issue is: What is my real time information need coming into the system in the first place?

No amount of system building will change my lack of such a need, will it? Or are you saying that the more I play around, the more I will discover what my needs actually are?


Thanks, Jeremy, for playing around with TipTop some more. What I was hoping you would start to see more clearly is exactly the latter. What should your needs be? Of course, as you might have already guessed based on my earlier posts, I am not obsessed with sticking to any terminology that much. The ultimate search and discovery experience will have significant amounts of real-time components. For example, in TipTop Shopping we let you chat with someone who purchased the very product you are looking at just a minute ago. Discussing the product with them at that very moment is likely to be more effective then, say, exchanging e-mails over the course of the next few days. I can tell you about a number of other such scenarios.


Yeah, whatever terminology gets used, do I really need to find someone that bought a product 1 minute ago?

When I bought my first digital camera, I got my best advice from someone who had bought theirs over a year prior.

I just don’t see what “1 minute ago” really buys you, nor how I would discover that by searching tweets. But I am of course completely open to changing my understanding. I could be completely naive and dense here, and everyone else sees it, when I can’t.

What do they say on Seinfeld? “It’s not you, it’s me” 🙂

So yes, please describe or give pointers to those scenarios.


Jeremy, you are right. I am not making the point that one excludes the other. For you the person who bought the camera a year ago might be most useful. For me it might be the one who just completed his research and is now ready to buy the one they like.

I find our conversation very interesting and would love to connect by e-mail or chat on the phone. You can reach me at shyam AT Daniel, Terrence, and anyone else who visits this blog is welcome to contact me by e-mail as well. I am a long-time member of the IR community and love to chat with the more innovative folks within it like each of you & your readers appear to be.


Comments are closed.