The search for The Answer

I had the privilege of hearing Google CEO Eric Schmidt speak at the Economic Club of Chicago last week. Matt McCall has a great recap of some of his main talking points.

Eric explained that the Internet has given rise to paradoxes that are new and difficult to understand.  For example, the Internet encourages openness and self-expression, like we see in blogs, while it simultaneously fosters “tribalism,” like hate groups that find each other online and isolate themselves from others.

He went on to talk about how in the future you will be able to buy a Google device that contains a search engine and all the content, so that you can get the answer to your question on a small computer that you carry with you.  Eric thought this would be particularly good for students.  Maybe it would.  In the future, everyone will access everything through Google.

In the same speech, Eric contrasted the old way in which culture was disseminated – through broadcast media, controlled from the top down, with the new way – in which user-generated content spreads like wildfire without any central control.

But Google’s algorithms, the procedures that determine what you see when you type into the search box, are that central point of control.  What you see first is determined by its popularity or “authority” and the algorithms drive towards giving you the “correct” answer.

This is the innovation the Google founders are most well known for.  PageRank, one of the factors Google uses to determine the order of search results, is the idea that links should influence relevance.  Prior to PageRank, most web search algorithms (except Kleinberg’s hubs and authorities algorithm) only considered a page’s contents, not how many other pages linked to it, or what the link said (more evidence for what the page is about).

The effect of PageRank is that the most popular answer is the one that will appear first.  This is good because it’s most likely to be the answer you wanted to see, because, the thought is, all the people on the web have voted by linking and determined that that’s what your search terms mean.

Good until you understand that people rarely browse beyond the first page of search results, not to mention that the ranking can be and is “optimized” artificially.

So, what about all the other answers?  What about all the content that doesn’t make it into the first page of Google results?  What about all of the content publishers who don’t pay companies to optimize their ranking?  What about all of the other factors that influence whether something is relevant to you?

PageRank is the new “old way” – just like the networks control what you see on TV, PageRank controls what you see on the internet in its search for “the answer.”

Perhaps this is the most troubling paradox.

Instead of searching for “the answer,” we should be empowering users to explore the landscape of answers.  Instead of driving to a single result, we should be showing the user the range of contrasting opinions, and giving them tools to discover the ones that make the most sense to them.  The future is not “the answer.”

Comments

[Dan Balzer]

Thank you for this insightful challenge. I was struck by your phrase, "Instead of searching for “the answer,” we should be empowering users to explore the landscape of answers."

Getting searchers to find THE ANSWER is a technique we use in our Live Internet Search Challenges (http://21cif.imsa.edu/tutorials/challenge/SC001/SC_directory.html) to help learners assess their own performance and discover more efficient ways to search. That is a core competency of the first phase of the Digital Information Fluency Model (http://21cif.imsa.edu/resources/difcore). The question that you raise here about relevancy fits most closely, I believe, with a key question of the third phase of the DIF model. How do searchers know when they are looking at relevant information? In our observations, searchers often skip over keywords in the snippets that could lead them to move relevant "answers".

How do we teach this skill?

Dan

[Watson]

Dan, this is an interesting problem. Sometimes we know the right questions to ask, other times we don't. Sometimes we don't even know we should be asking. Research starts before the initial question: you have to know you need to know something and you have to decide what to ask for.

In an age when technologies and tools are changing so quickly, it makes sense to me to focus both on teaching skills that get the job done (how to use search engines) but also on skills that aren't dependent on the tools at all.

Searching as we know it today will surely change, and we hope to usher in that change. We think people should be spending their time doing more important things than deciding what terms to select for their queries.

Skills like what question should we be asking in the first place, and then secondly what questions we should be asking about the information in front of us will be with us forever. Is the information biased? How does it compare with other sources? Is the source authoritative? What have other people said?

To your specific question about search engines, snippets, and relevance, I think one important tip is to envision the web page or document that would contain the answer and then think of what terms would be used in that document and not any other. Terms that occur in the document you're looking for but not any other are good "discriminators" and help get rid of irrelevant hits. These terms could come from the snippet or from content the user finds along the way, as they become more familiar with the language used by authors. The more terms like this a user uses, the more precise the results become. This is what systems like Watson do to narrow a search.

Of course, if we have our way, you won't need to teach people to do this at all -- their computer will do it for them.

[Carl Carpenter]

You've made a beautiful point here, Dr. Jay. The final aggregation, the collapse of the information wave into time, into a conscious mind, rightfully occurs in the individual, and this can perhaps best be accomplished by sampling and aggregating from among a machine-generated landscape of answers.

However, I respectfully disagree with your contention that 'the future is not the answer.' The ultimate interface to search is conversational. That means the machine must perform the final aggregation in order to produce the singular answer it ultimately articulates.

So long as the machine is a tool, final aggregation need take place only in the human individual. But once the machine can speak as a separate individual, then there are two final aggregations occurring, and in that case, if the machine is intelligent enough and the answer is perfect, the machine's final aggregation becomes the individual's final aggregation. This would seem to represent optimal efficiency.

For the time being it is a tremendous stride forward to be able to perform our individual final aggregations with the assistance of a tool like Watson. But when Watson one day individuates, when it speaks as an individual intelligence, then our efficiency of intelligence will reach a new level, for then proof will be offered without our effort. That would seem to be the ultimate future, a future where the machine determines the answer we virtually always accept.

I just discovered Watson today. I'm going to download it now, and look forward to trying it out. BTW, have you considered an Ajax version of Watson for bloggers?

[micfo]

The future is not “the answer.” -I foresee the future search engine is more interactive and going to provide you only one answer and that will be the best answer because it is going to be personalized with your personal search habit.

-Bob

[govokinolij]

Hello

Looks good! Very useful, good stuff. Good resources here. Thanks much!

Bye

[pleadly]

The Pleasing text and design!
http://danuegonax.com
I Will be back!

[pleadly]

Very good web forum, great work and thank you for your service.
http://srubibablo.com
Well, thanks!

[honda-radio]

Thx! :)

[Oxydaytwesest]

I like you site - blogs.intellext.com

Glad to show you path to
no fees
interesting
place.
Try to see it
online exam training and pre-tests http://www.wwon.net
is there for free.

[Indishinnorry]

ball prescribed, notional.