Monday, November 10, 2008

Tasks, not search, at DEMOfall08

Philipp Lenssen points to a video of a DEMOfall08 panel on "Where the Web is Going" that included Peter Norvig from Google and Prabhakar Raghavan from Yahoo.

What is notable in the talk is that (starting at 12:32) Prabhakar and Peter agree that, rather than supporting only one search at a time, search engines will soon focus on helping people get a bigger task done.

Prabhakar says:
I think the next step that faces us ... is divining the intent of what people are doing [and] fulfilling their tasks, getting what they need to get done.

People intrinsically don't want to search. People don't come to work every day saying I need to search ... They want to run their lives.

The notion of a [mere] retrieval engine as the ultimate [tool] is incredibly limiting. We have to get much further along to task completion and fulfillment.

The current world of search engines [are] stateless ... In the act of completing a task -- booking a vacation, finding a job -- you spend hours and days start to end. In the process, you make repeated invocations of search engines ... And all the while the search engine has no state about you. It doesn't recognize that you are in the midst of a task.

What does it take to recognize an intent and synthesize an experience that satisfies an intent? So, if it is a vacation you are planning, it should say ... here is a package I recommend that is based on your budget, the fact that you have two kids, that you don't want to go to too many museums. That's the future we have to get to.
Peter says:
We have to get a lot farther in saying what is it that the user means both in ... tasks where there is a clear intent ... and [even] more so in cases where it is exploratory.

We see this all the time that the user has some area he wants to figure out -- let's say a medical problem -- and the user starts out by kind of floundering around not sure what to talk about and then he reads some document and then he starts to learn the lingo. And, now they say, I don't say funny red splotch, I use this medical term and now I'm on the right track.

We have to accelerate that process ... not make the user do all the work.
There is an interesting shift here from a model where each search is independent to one where a search engine may expect searchers to do multiple searches when trying to accomplish their tasks.

That new model could take the form of search as a dialogue (a back-and-forth with the search engine focused on helping you understand what information is out there), personalized search (re-ranking your results based on your past actions, interests, and goals), or recommender systems (helping you discover interesting things you might not know exist using what people like you found interesting). Most likely, I would expect, it would require a combination of all three.


Daniel Lemire said...


How much is this about getting Google Docs, Google Mail, Google Search and all these tools working together so that Google can own the world? (Disclaimer: I use all their products and I am not anti-Google. But I try to remain critical of anyone with too much power.)

Getting back to the IT or computer science problem lurking...

This is a general problem. For example, Amazon does not want you to find out about all products containing the string "dog". They want you to buy as many books as possible. Yet, to help you get there, they provide simple tools.

I am a bit afraid that Google could go the "Microsoft Office" way... and part with the "Unix" approach we have used so far. I mean, by going beyond search, are we just going to bundle new applications together or are we going to push search further? Bundling could be of great value to Google, but not so much to us.

The dialogue angle is interesting. It has been tried, but I guess it failed in various ways. Mostly, people are busy and they don't want to be chatting with their search engine or word processor unless it brings about tremendous value. Can Google pull it out?

Of course, what we would be nice is that give a search result that points to a given algorithm, Google points to an implementation, and they to a library, and then to the document that goes along with it. In effect, serving as a super smart personal assistant.

However, good luck predicting what I am going to do next once I see the search result.

I do not know how others organize their time and manage their projects, but I use greedy algorithms. To quote myself:

I seriously doubt that anyone can manage his time better than with a greedy algorithm.

hardtke said...

Many of the concepts and goals envisioned by Prabhakar and Peter are implemented by our Surf Canyon technology. In particular, we modify the search experience on Google, Yahoo and Live Search such that it is task-based and context aware rather than single query based. We call it real-time personalization based on implicit relevance feedback, but it could also be described as the persistent task-based search engine they discuss.

As part of our development, we have done many experiments to see if this concept works in practice. We put together a paper (blog post full paper) with a description of the experiments and our findings. We find clear evidence that users prefer search rankings that account for the a user's task, recent actions, and recent searches. If we interleave our context-based search results with the original Google results, we find that users are 30-40 percent more likely to click on the personalized results. I doubt any advances in stateless search technology can match these gains -- user specific, task-based and persistent is definitely the future.

Anonymous said...

All this talk of tasks and personalization reminds me of Microsoft Bob. :-)

Anonymous said...

...but in all seriousness, I wonder whether Google will have the guts to sacrifice its "clean" interface in support of task-based information seeking.

It was one thing to stick to that notoriously sparse interface when searches are stateless. But adding state means adding information. Maybe not click-able information -- certainly there is an ongoing battle about search-as-dialogue vs. search-as-implicit-short-term-personalization. But information nonetheless. Information that at least shows you what you've saved, or gives some sort of awareness of the search-state you are in.. so that the user doesn't get too frustrated by having to do all the work themselves in keeping track of where they are in their overall task.

But in order to show that additional information, the interface will have to become less sparse. So my question is whether Google will be willing to do that.

Or does someone have a counterargument as to why you don't need to show the user any additional information, no matter what form, in a state-rich, task-based search?

Greg Linden said...

Hi, Jeremy. Sure, I could see ways of using state that don't require adding much to the interface.

Google Personalized Web Search is a pretty good example of this already. They reorder and annotate the results based on state, but make few changes to the interface.

I also think there is a lot we can do further down the page. On navigation queries, for example, the top result is all that matters. What we put below that result could focus on information exploration more than it currently does.

While perhaps it could be done without major interface changes, I wouldn't say that it should be done that way. But, I wouldn't look to Google for major UI innovation. I think smaller players have much less at risk if they were to mess up while trying to change the tried-and-true look-and-feel.

Anonymous said...

Google Personalized Web Search is a pretty good example of this already. They reorder and annotate the results based on state, but make few changes to the interface.

I agree with you that personalized search allows one few changes to the interface.

I'm talking more about the "booking a vacation" task example. In that case, your state includes things like a checklist of whether you've got the (1) flight, (2) hotel, (3) car rental, (4) reservations at Chez Fancypants, (5) Tickets to the hit show on broadway, etc.

Additionally, since vacations are not linear, unbendable death marches, most people usually have secondary and tertiary plans and backup plans. For example, in case the star singer at the hit show is sick that evening, you might want to know that there is a museum nearby. In case it rains on the day that you were planning on walking the central park, you might want to know that there are certain cafes between the park and that new science exhibit that you had considered.

Planning your trip, and planning contingencies into your trip, so that you don't have to scramble in the moment when something doesn't work out, involves a more intricate, deeper interface. At the very least, it involves an interface that lets you check off all your main tasks -- that gives you a conscious awareness of the state of the overall task and system.

So personalization is not really the issue here. Neither is search-as-dialogue. It's the fact that you have a broader and/or deeper search need that involves keeping track of state, solving subtasks, and being aware of alternatives.

And even keeping track of that todo list is more complicated than Google usually likes to be.