The Evolution of Search


“Google began
as a research project in 1996.” “This is a look
at how search has evolved.” Gomes: Our goal is actually
to make improvements to Search that just answer
the user’s information need. Get them to their answer
faster and faster. So that there’s almost
a seamless connection between their thoughts and their
information needs and the search results
that they find. Singhal: Well, Google was
started based upon algorithms that Larry and Sergey developed
in Stanford called the PageRank algorithm. And they used that algorithm
to indeed build a very novel way
of searching the web. Gomes: What was happening
at that point was there was this huge explosion
of content in the web, a bigger explosion
of information than had ever happened before. And it was getting
increasingly hard to find the piece of content
you wanted. “Adwords” Mayer: In the beginning, we didn’t have
any advertisements at all. And as we went
to add advertisements, it was very important to us
that those ads be as relevant to the search as
the search results themselves. It was also very important to us
that they be distinguished from the search results. Gomes: There was a clear
separation between ads and Search from the very
earliest times. Search had one goal
and one goal alone, to provide the most relevant
information for the user in the fastest time possible. “Universal Results” Mayer: In 1999 and 2000,
we had a search engine that worked wonderfully, and it worked wonderfully
for web pages. One of the things we saw was, as Google got better and better, users expected more
and more from it. They didn’t want
just web pages. They wanted the best possible
information available, be that a picture or a book. And so we started looking
at how we could search new and other forms of content. And Image Search was
the first of those because we know that a picture
is worth 1,000 words, and there were a lot
of times when people would say, you know, what is turquoise? You know, we got a search,
“what is turquoise?” And there’s no way
to answer that question without a picture. Singhal:
When September 11th happened, we as Google were
failing our users. Our users were searching
for “New York Twin Towers,” and our results had
nothing relevant, nothing related
to the sad events of the day because our index was crawled
a month earlier, and, of course, there was
no news in that index. So we placed links to all the
news organizations like CNN right on our front page saying
please visit those sites to get the news of the day, because our search is
failing you. My friend Krishna and I were
attending a conference at the time, and Krishna started
thinking about the problem, saying, “If we could
crawl news quickly, “and we can provide
multiple points of view “about the same story
to our users, wouldn’t it be amazing?” That was the birth
of Google News specialized search service. Mayer: Well, in 2002, one
of the trends we started to see was that the web had become
a lot more rich. More images, more video,
different kinds of content. And we started to realize that
our users expected Google to be able to find something
if it existed on the web. They didn’t care if it was text
or a web page or news, they wanted it all
in one place. And so we came up with this
notion of Universal Search, the idea that you could just
go to Google, and no matter what type
of content it was, we could find it. One of the challenges that came
up in Universal Search is that we were really comparing
apples and oranges. Singhal: You can imagine
these apples are web pages, and you can imagine an orange
would be an image. When we look at ranking images, we know what the aspect ratio
of an image is, how big is the image,
how many pixels this image is. Is it a black and white image
or a color image? And all these signals are
only relevant to images, but are not relevant
to web pages. And that’s why Universal Search
was such a hard task when we did it, because the
science was not fully developed. Mayer: We basically ended up
putting things either on the top of the page,
on the bottom of the page, or somewhere in the middle because we didn’t have
a finer-grained way of looking
at the relevance, especially across
different media types. Singhal: And over the years, we developed our science
tremendously, and today we are beginning
to place several kinds of information
in multiple positions on our result page as our algorithms get better
and better. “Quick Answers” Gomes: Our goal is to make it
so that the improvement we make is so much what you wanted and fits so cleanly
into the flow of what you are looking for that you almost don’t notice
that it’s happened. And looking back at it, it seems
obvious that that’s the way it should have always been
in the first place. Menzel: When you do need a
specific bit of information, Google tries to provide you
exactly that using our Quick Answers. For example,
take sport scores. You kind of want to know
what the score is right now. You want to know how tall is
the Empire State Building? Wright: We want users
to come to Google and get their information
as quickly as they possibly can. And with Instant,
you don’t even have to type in your full thought. You don’t have to hit enter. You can type in
something like, “bike h,” and we’ll just show you
the results right there before you’ve even finished
your thoughts. “The Future of Search” Menzel:
We are pushing the boundaries of how you actually
fundamentally interact with the search engine itself. With Search by Image,
you can actually use that image as an input for the search. Singhal: The truth is that
our users need much more complex answers. My dream has always been to
build the “Star Trek” computer. And in my ideal world, I would be able to walk up
to a computer and say, “Hey, what is the best time
for me to sow seeds in India, given that monsoon
was early this year?” And once we can answer
that question, which we don’t today, people will be
looking for answers to even more
complex questions. These are all genuine
information needs. Genuine questions that,
if we, Google, can answer, our users would become
more knowledgeable, and they would be more satisfied
in their quest for knowledge.

Comments 98

Leave a Reply

Your email address will not be published. Required fields are marked *