Excerpts From Google Press Day 2007
Originally Posted at Google Blogoscoped
The world's most popular search engine in getting larger by the minute. For more perspective on the Google System and its scalability, let's review some 2007 outtakes from key Google execs.
CEO Eric Schmidt:
“Search was, and is, and I suspect will be for many years, the killer application,” Eric says. “We have more engineers working at Google on search than we do anything else.” And: “As the web gets bigger, you need a bigger index ... It’s like looking for more needles in larger haystacks.”
Personal search is the next big phenomena, Eric says. The best search is a personal search – one that we arrange, we control, etc. This is going to become the theme of Google as they move forward, Eric proclaims, stating that (and this surprises Eric) iGoogle is exploding in its use. Eric likens iGoogle to a ring-tone as people love to personalize it.
Eric continues to say that Google likes to tackle things with the help of partners, sharing revenues, and that there are lots of challenges ahead; growth, society and the future. “What are the next billion people who come online going to do? Are they going to democra
tize the web? Are they going to bring more languages? I don’t know, but I know they’re going to come.” Eric mentions a Venezuelan TV station that survived by coming online... and using YouTube. This was an important event for the people of Venezuela.
Eric talks about how soon, it will be possible to walk up to a computer and make it yours: the “Be Mine” command. Securely, safely, without any question of theft. And then, when you’re done, you’ll say “Be Gone” and you’re logged out.
Marissa Mayer:
So far, she spoke about the core principles and components of search – the four answers which create the entire experience that you have on Google:
1. Comprehensiveness
2. Relevance - Which goes first? Apples, oranges or grapes? This is what they’re doing with websites, Marissa says: it’s a question of ranking and ordering the results, the question being how can they rank the results in the most relevant way possible?
3. Speed - Marissa says they’re constantly working on making Google faster and faster.
4. User Experience - Google asks themselves, “Do the results make sense? Is it easy or hard to understand the answers and the layout of the pages?”
With that in mind, Marissa says she wants to take a look at Google’s past, present and future. In the early days, around 1998, the world wide web would fit on one slide. This year is Marissa’s 8th year at Google, and they now have 10s of billions of slides, grown by more than a factor of 1000 in eight years. Back then, you could organize all websites in a manual directory, as Yahoo did, but as the web exploded search became a necessary tool.
Google is constantly working to crawl and include more and more content, Marissa says. In June 2000, a year after Marissa joined, it was the time of GIGA Google – 1 billion web pages crawled. Next search engine had 600 million. As you include more pages, relevance bec
omes a problem. In the early days, users got used to hunting and pecking at results. Google realized their job was to provide the best result first, and that if they did their job right, the user shouldn’t have to click “next” or even scroll. This lead to the “I’m Feeling Lucky” button, which has by now been translated into many languages.
Basically Marissa says Google realized they only had a few users, but they started to see the user base grow due to people getting better search results – which made it spread across the US and internationally.
Marissa says there are a lot of speed advances in the past 20 years. Two decades ago, you’d go to the library, look things up, get journals out and find your answer. Many questions would go unanswered because it wasn’t worth spending hours researching some questions. Then, you could email friends and sometimes they had the right answers. But with internet search, people could get answers within a second. But at some point, Google will be up against the constants of physics – though they’re aiming for light speed searches.
In an early Q&A that followed, Marissa was asked about the size of Google’s index and told the press it is “10s of billions of pages” and “three times bigger than next nearest rival.” (I wonder what Yahoo has to say on that...) Also, someone mentions how Jason Calacanis
thinks the internet is polluted, and that we need a human edited search engine. Marissa replies that when the ’net is large & polluted, you need more sophisticated means to help people find information. Ideal, she says, is a “blend of both” an algorithmic search engine and a human-edited one.
After lunch break, the presentation continues, and Marissa shows screenshots of the evolution of the Google homepage over the last eight years.
Apparently the first version of the homepage was so simple in the because co-founder Sergey Brin didn’t know much HTML. Marrissa adds that her family gives her a hard time over the look of the Google search result pages, as they say, “It looks the same! What do you do at work?”
Marissa goes on to speak about improvements to web search. One example are Google’s alternate queries; search for ABC Survivor and Google understands the semantics of this query and suggests “CBS Survivor” as well. Marissa also mentions that that Google serves different results for different locations - e.g. Cote D’or in Australia will return the chocolate brand, but in France they’ll localize the results to the region in France. (And in Belgium, they’ll return both.) Speaking on the future of search, Marissa mentions the Cross Language Information Retrieval designed for search in one language, to find documents from other languages,
translated back to your preferred language on the fly, and she reiterates the universal search concept, saying that “information silos” – Image search, Blog search, News, Video etc. – are united to break down search barriers.
Marissa also touches the subject of personalization. “I think the really powerful part about all this is that we can take information from Web History and iGoogle so that we can create a search engine of the future.”
In a second Q&A with Marissa, someone asks about potential copyright issues when the Cross Language Information Retrieval is republishing full pages for its translation service. Marissa admits that they translate search phrase, titles, descriptions and entire web pages, but that they believe this is legal with regards to copyright law. Another question mentions that Wikipedia is heavily featured in Google results; does Google have discussions about reliability and authenticity of any results, e.g. should the BBC website rank more highly than someone’s blog? Marissa answers that PageRank is user-driven by linking to pages. Links feed into page rank, anchor text, etc., so those types of links to
Wikipedia happen because people like the content and link to Wikipedia pages.
Questioned on googlebombing, Marissa says that links on blogs often look like googlebombs when they’re not. But she also says that googlebombs often become the “right result” in itself because people searching for e.g. “miserable failure” actually want to find a page on George Bush! When asked how much Google invests in video image recognition, Marissa states that voice-to-text recognition is more important for video search because it’s further along than image recognition.
Urs Hölzle (Google Senior Vice President, Operations) is responsible for energy usage and carbon dioxide emissions. Google has a focus on cutting their emissions and “working to raise industry standards,” but the average Google data center catering to the internet’s many users and services wastes power. Computers use a lot of energy, which turns to heat, which needs cooling, which needs more energy. Urs says in a typical data center, up to 65% of the energy is lost – only 35% reaches the computers!
Urs Hölzle:
To tackle some of the issues, Google is committing to being carbon neutral. They already use evaporative cooling, Urs says, which according to him results in data centers that use “50% less energy than standard industry data centers.” Goog
le also runs a shuttle service on Bio diesel in the San Francisco area, gives free bikes to employees, uses video conferencing to reduce travel, and has solar panels installed at their headquarters. Urs says that Google is “offsetting what is left,” which he says isn’t always the best, but a “good compromise.” Urs’ speech ties into yesterday’s announcement of Google Recharge, as well as the recent Climate Savers Initiate.
Also on stage speaking now is the French photographer Yann Arthus-Bertrand, who specializes in aerial photography and whose works have “both political and aesthetic connotations,” as Wikipedia writes. Yann talks about the people who are
going to suffer with global warming. Jim Walker of the Climate Group – which Google joined a couple years back – also makes an appearance, stating that Google’s announcements are “encouraging” and that they “reinforce their vision” on environmental issues.