SEARCH ENGINES: David vs. Google
How a Mathematician from New Jersey wants to Overcome the World's biggest Search engine: A Trip into the World of Algorithms
Apostolos Gerasoulis is a mathematician and normally deals with solutions for problems the world can afford to wait for. Recently, however, in the evening, when he and his wife sit in front of the TV, he becomes a little nervous and excuses himself for a few minutes. His wife knows that it will take a bit longer when he secludes himself in his living room in front of his Dell Inspiron laptop. That's because he checks what was on the world's mind on that particular day.
Apostolos Gerasoulis is the co-inventor of America's forth largest Internet search engine, Ask Jeeves. Every day, ten million search phrases are entered here, amounting to ten million questions and requests, from news and products to diagnoses and names and obviously naked women, porn, strippers, orgies and Pam Anderson videos. But they also ask about the meaning of life, about God and the devil, both of which, however, are far behind Britney Spears, which alone among world celebrities has remained on top for several years now and for whose name Apostolos has seen hundreds of spellings.
He is sitting in front of his computer, wearing shorts. A 54 year-old male with a little weight problem and straggly hair in his living room, in a big house under old oak trees at the end of a cul-de-sac, in a New Jersey suburb. His nine-year-old son sometimes asks whether it's possible to become something big when you already have everything as a kid. Two years ago, Apostolos had the two-floor mansion with its column portico built over a small cottage from the 1930s. The living room is all mahogany. When he sits here, he can see an old golf course which laps against his porch in gently rolling waves. At midnight, everything is quiet. He focuses on the silent cacophony of questions people entrust to his engine. From above, his wife says that she's going to bed.
Of course, he can't deal with each little detail. Apostolos Gerasoulis is Professor of Applied Mathematics. He believes, that in numbers lies truth. On his screen, words amount to statistics, and the statistics amount to a history of the past day. On this August evening, as usually, Iraq and the U.S. are among the top ten of the country list. Today however, two new voices made themselves heard in this huge computer choir, one of which, it seems they're against the war in Iraq, write "Grieving Moms" and "Out of Iraq", the other group, obviously supporters of the war, asks for "Iraq despite deaths" and "Northern Offensive". "You can witness trends that some day might have important resultsâ, says Apostolos. This happens often: He can often determine the winners of TV talent shows in advance from the number of search requests. Cardinal Ratzinger was also searched more often before he was elected pope than his competitors. Sometimes, he can see how excitement or nervousness influences his statistics, "Tsunami" or "New Orleans", or compassion, "Red Cross" for example was a search request entered much more in the U.S. after the hurricane disaster than after the tsunami in Asia. "Is there a correlation between the distance to an event and the amount of compassion?" he asks. "Can we extrapolate from these figures how big the psychological shock was?"
Rarely are there days on which no new word occurs. This is when everyday life and all its routines enter the web, on Friday it's movie tips, on Saturday shopping requests, on Sunday homework. Turkey recipes shortly before Thanksgiving, then the word "turkey" disappears again and resurfaces for a while right before Christmas. Apostolos calls this a seasonal request, just like "love", which has a short-lived peak right around Valentine's Day and then plummets quickly. "Sometimes it is as if I can sense the world's feelings, which is a heavy load, too. What if we would answer the wrong way to requests like "love" or "hurricane"? Apostolos Gerasoulis believes that there are right answers to all questions.
It's all about a kind of hunger which is still far from being satiated. Just like kids, most adults have their heads full of questions, but there's nobody at hand who knows more than them, so they stopped asking and learned to postpone their questions. Until Google appeared, the possibility to get an answer from the immense collection of human knowledge, from this Library of Babel, as the Argentinean writer Jorge Luis Borges dreamed, a collection which contains everything, all letters in every possible combination, all meaning and nonsense in all languages. In Borges' story, people in the library get old without finding an answer to what they searched for.
Today, it only takes fractions of a second until a search engine has found thousands (more or less fitting) texts to respond to the questions humans are occupied with at a certain moment. More questions are the results, this childlike enjoyment of keeping asking questions, to follow references, until one looses track and ends up somewhere in Absurdistan. To google â this verb has become such a part of everyday life that it was added to the official German Duden dictionary. And when Google was down for a quarter of an hour in May due to server problems, the news agencies found that event newsworthy. The dependency of this supplier of answers hardly anybody even knew a mere five years ago is immense. When Google is down, it is as if the world's water supply has been turned of for 15 minutes.
What a market! On one side human curiosity. On the other, the Internet. In-between are search engines which try to provide ever more accurate answers. Ask Jeeves is the smallest of these search programs with a market share of about six percent in the U.S. and in the U.K. Its answers have reached Google's accuracy. There are industry rumors that say that Ask Jeeves uses the most innovative technology. Ask Jeeves is surpassed by Yahoo and Microsoft. Yahoo entered the market in early 2004 with its own search engine, then Microsoft got active, too, and all fear that the monopolist could steer the masses to its own search engine MSN by the power of its operating system. Above all, there is Google, 300 million users, two billion requests a month. Worldwide, Google manages approximately half of all search requests, in Germany even more than 80 percent. "The big G", says Apostolos. G like God. Or like Goliath. He is David.
The corporate vision of Google is "to organize the world's information and make it generally available." Apostolos Gerasoulis' vision is to deliver better answers than Google. He believes that he could take market share from Google or might even end its dominance.
Everybody dreams that. If only one could find a magic spell which delivers better answers. Just like Google did when it made all other search engines look old, whether they were called Alta Vista, Excite or Fireball. Search engines which fell behind but are still somewhere in the Internet, at remote locations, far from everyday data streams, their best parts taken out and reused like car wrecks along a desert road.
It could happen very fast, says Apostolos. "Everything you need is a whole lot of computers and extremely fast internet connections. And the magic formula. Lots of people are looking for it right now. In the last two years, Microsoft has spent hundreds of millions of dollars to enter the game. Daily, I receive an email saying that somebody has found the best search algorithm."
The U.S. industry expert Charles Ferguson estimates that half a trillion dollars could be made in this market in the coming decades. Soon, the main focus is not only going to be web pages, but everything: TV shows, book texts, databases, private stuff, recordings of phone conversations.
Of course, the profit comes from advertisement. If you have the entire world's attention, you can sell a whole lot. For search engines, this happens via so-called ad words, bought words that are displayed in the column to the right of the search results and match the search request. For the advertising industry, it must be a dream. Just as if people in the street would only see ad posters about the things they currently are thinking about. Google's advertising revenue is already higher than that of the New York Times. In 2005, advertising money spent in the Internet has increased by 20 percent. Ask Jeeves made a profit of 261 million dollars last year, Google of three billions. Google's share value is 80 billion dollars at the moment, more than DaimlerChrysler's. Apostolos Gerasoulis says he could stop afford to stop to work, if it was only about the money. But now it's about the world. He says he's a happy man. A little ago, a few hundred people were interested in his algorithms. Now they influence the lives of millions of people.
He of all people! A mathematician! Always somewhere else with his head. His favorite word is "abstraction", the grand theme of Apostolos Gerasoulis. When he was a little kid, he already drifted off into the world of numbers, because the real world, a tiny, poor mountain village in Greece, was too boring for him. His father had a hardware store. At first, he was a pretty bad student, until he discovered the principle of abstraction in the stories by Dostoyevsky. Best math student in high school, best math student in all universities, first in Greece, later in the U.S. He really is Professor of Applied Mathematics at the Rutgers University, the public University of New Jersey, specializing in algorithms. Algorithms are plans of processes which describe how you get from A to B. This can be formulas or recipes. His algorithms mostly dealt with programs for super computers. A few other scientists were able to understand them, and sometimes it felt to him as if they could be of no relevance to the world. Until he confronted his students with a problem in September 1998: How can the documents in the internet concerning one topic be put into an order that makes sense?
The homework became a formula which then became a search engine a venture capitalist became interested in. That was in 1999 and was the origin of the search engine Teoma. In 2001, the company â Apostolos, his vice president Tao Yang, eight employees and their search engine â was bought by Ask Jeeves for 4.5 million dollars. Ask Jeeves is named after the eccentric butler from the stories by P. G. Wodehouse. Today, the company has offices in Japan, China, Italy, Spain, England and Ireland. By the end of the year, German-language search functions are supposed to be available as well. Last summer, the Internet group IAC acquired Ask Jeeves for 2.3 billion dollars. IAC is owned by Barry Diller who is regarded as a visionary. He used to be the head of Paramount, later he established a fourth big TV broadcast channel in the U.S. when he founded Fox News. The 63 year old Diller is part of a media revolution in which not only the traditional mass media make their voices heard, but the line between authors and readers gets more and more unclear. In the Internet, everybody can publish and everybody can decide by him- or herself what he/she wants to read. The only thing between author and reader are the search engines. They have a significant influence on which messages are heard or paid attention to. "The geeks have taken over", says Apostolos.
He has the facial expression of a child, always slightly astonished, sometimes almost happy. In the office, he wears kaki pants and worn-out t-shirts which show his first company's logo, Teoma. If a coffee stain gets added during lunchtime, the stain stays visible for the rest of the day. This is the geek dress code. First they automated the production processes, now they automate the dissemination of information. People used to ask: Which newspaper are you reading? The answer revealed quite a bit about preferences, political affiliation and education level of the reader. Pretty soon the question might be: Which program do you use to get information from the World Wide Web? Journalists probably will remain suppliers, supplemented and corrected by private web logs. 100,000 of these Internet columns already exist in Germany alone. The composition of the daily information menu, however, is done much more efficiently by computers. Google News already delivers news compiled automatically by a program which evaluates 4,500 sources.
Thus, the algorithms deal with language, which normally is the domain of human beings, and which just a little while ago seemed to be wild and incalculable. Now, day in and day out, crawlers crawl through the documents in the Internet, programs which scan every page and go from there to the next page. Anybody who has a home page can see that, too. When the google search engine has read a page, it leaves a note: "Visit Google Bot". By doing so, the world's knowledge is being recorded by super computers whose storage capacity is measured in terabytes. Teras in Greek means "monster".
Here they are sitting in a brass mirror fronted office building entangled by highways in the middle of nowhere in New Jersey. Cut off from the rest of the world on a traffic island which can only be left via car. Approximately fifty programmers work on the fifth floor in blue cubicles. The visitor passes name tags with Indian or Chinese sounding names, migrants in cyberspace. If you stick your head into one of the cubicles, they are slightly shy, their handshake is a little hesitant, their demeanor a little uncertain, which might be a character trait from their Asian homelands or might just be typical for scientists or somebody in the rural hinterland. They are from villages and small towns, two thirds of them from Asia, a fifth from Europe. Nothing has prepared them for reprogramming the world's information flow, for their work of public interest. Most of them came to the U.S. in their mid-twenties in order to finish their university studies here. Their English is not great or difficult to understand, but they all write on a great text in C++. C++ is one of the biggest programming languages and is used primarily when huge amounts of data are involved.
C++ is constantly being enhanced, "the language becomes more efficient every year", says Wei Wang, who has written the main part of the Ask Jeeves program. He constantly learns new vocabulary which has been newly developed in order to be able to keep up with ever faster computers. Computer languages are a mixture of mathematic formulas and English, words brought into a strict logic. Commands which were formulated by humans and are being executed by machines. As a lay person, you think you can understand sentence fractures: IF my_document CONTAINS b THEN RETURN 0; IF my_document CONTAINS a THEN results.add(my_document).
Every programmer has his or her own style. There are many different options to say what you want. Many create a plan in advance, others simply start writing, some have a long and excessive style, others are very concise. Experienced people can see from a program which colleague wrote it. One of the employees even says that a well-written program is just like a good book, astonishing in its clarity and intelligence.
Wei Wang is known to be slow author. His programs are known to be straight-forward, direct. He achieves in one step what others need several steps for. He is also known for his attention to potential sources of errors, even before they appear. He calls this "brain debugging". It means to take into account all possibilities. His wife always tells him that he plans his vacations in a similar manner.
He taught himself C++ when he was still at Zhejiang University south of Shanghai, where he studied computer sciences. In summer 1999, he came to New Jersey, still a narrow-shouldered twenty-something young man. He took Apostolos Gerasoulis' class. During the first year, he hardly said a word and never did a presentation. In C++, however, nobody could say what they mean as correctly as him, as his professor noted soon.
Information Retrieval, the method of filtering a certain text out of a huge mass, is not new. For a long time, it was especially important for working in libraries. Since the 1950s, the discipline became stagnant. Then, the Internet grew in leaps and bounds, this parallel universe, of which nobody knows for sure how big it is. In recent times, Google captures some 30 billion documents, but the biggest part of the web remains unknown territory, a wilderness consisting of closed databases, company networks, remote home pages without links to the exterior world. The question, therefore, was: Who is capable of mapping all this?
At the beginning, there were lists, often created manually. Yahoo was founded in 1994 as a link collection of two students at Stanford University. Alta Vista was the first well-known search engine. Its algorithm was based on matching the search phrase with a document:
IF QUERY CONTAINS "Disney" THEN FIND DOCUMENTS CONTAINING "Disney".
The complicated part about a search engine is not to find texts in the internet which have something to do with the search request. The magic word is ranking, the order, in which the user is presented the answers to his request. The ranking algorithm is the most important part of a search engine. On average, users only click on two documents out of the thousands offered to them after a search request. A page appearing as the 52nd search result receives about the same attention as a flyer given out in a shopping mall.
Early Alta Vista's ranking algorithm evaluated the documents mainly according to the number of times a search word appears in a text. People started to write a certain word x times on their home page. Sometimes, they included words on their page which had nothing to do with its real content, in order to attract more visitors or customers to their site. If you wanted Disney, you got porn. If you looked for "God" you got the home page of a Chicago hot dog stand which used the advertisement slogan: "The god damn best Hotdogs in Chicago".
Apostolos Gerasoulis then was mainly interested in search engines because they used linear algorithms, his main area of expertise. During that time, he headed a tem at Rutgers University. He presented two essays to his students which were the topic of hot discussions amid experts. They dealt with the question which ranking method should be used to sort the documents in the Internet the most efficiently. One of these essays came from Stanford. Its authors were two young computer scientists, Sergey Brin and Larry Page.
Their idea was that the Internet itself makes a statement concerning how important a document is, by the number of links referring to it. The weight of these links themselves could then be determined by the number of links referring to them, and so on. Nowadays, some 50,000 links refer to a national newspaper. The private home page of a blogger might be referred to by three of his buddies. However, if he manages to write very interesting texts about a certain topic, just like the young Iraqi Salam Pax, who reported from Baghdad during the war and whose Internet diary became famous, it can happen that more and increasingly important pages refer to the authors page, resulting in the fact that he might appear further up on the list than professional journalists. The search engine developed by Brin and Page was supposed to function accordingly.
Apostolos Gerasoulis found the approach to be a good one, but not quite as elegant as the second idea. Its author is John Kleinberg, an American mathematician. Kleinberg went even a step further. He proposed not to count all links, but only those stemming from pages which concern the topic of the search request. The problem of this approach: long processing times. With the first approach, the degree of networking could be determined independently from the current search request. With the second approach, the processing was dependent on the search phrase and had to be determined anew for each input. This could take hours. For Apostolos Gerasoulis, it was a mathematical challenge to find a formula which could shorten the processing time.
In 1999, he used to sit in a windowless computer lab of the university almost every night, together with ten students from his team. Wei wrote a program which took one and a half hours in order to provide search results. Apostolos worked on a mathematical solution to accelerating the Kleinberg idea. And while he was still brooding over his fast matrix multiplications, the two PH.D. students from Stanford already put their engine into the web. They called it Google, a play on the word Googol, which describes the number 10100.
In August of 1999, they had shortened the search to mere seconds. Apostolos Gerasoulis recommended his students to interrupt their study work and patented his program. For him, it was like leaving the abstract sphere and landing back in the real world.
The Ask Jeeves formula is located in a data center in Boston and on server in the offices, translated to binary code, the mother tongue of all computers. It is protected by four passwords and not connected to any external data connections. Nobody has access to the entire algorithm and it would probably be impossible to know it in its entirety: it is simply too long. A human being would take months to read it all.
Constantly, some chapters are rewritten. Approximately twice a year, a new idea is integrated. The implementation of new programming parts resembles open-heart surgery. Ask Jeeves employees operate in the early morning hours, when the number of search requests is smallest. The whole office congregates in order to see how the engine's answers change. On such days, Apostolos is nervous when he gets to work. "The algorithm reacts in its own way."
A big search engine's algorithm can do quite a lot. It has to differentiate spam, undesired advertisement, from information. It has to recognize pages which only pretend to be linked particularly well. It has to uncover "link farms", connections which were only created in order to make a page appear particularly important. It has to recognize if a page is really about a certain topic or really only wants to sell Viagra. It has to separate news pages from diaries and sort out duplicates.
Some search engines even know their users: If you download the Google toolbar to your desktop or extend the settings on Google's web page, you agree that your requests are saved under your computer's address and might be assed on to the authorities, for example in case of a suspicion of terrorist activities.
A country's laws are encoded into the program. When Google China was founded in 2002, the company agreed to Chinese censorship regulations and removed content critical of the regime from the index. In Germany, some porn sites are not displayed because they are on the black list of the Freiwillige Selbstkontrolle (voluntary parental control association regulating adult content) for multimedia service providers. One of these service providers is Google. When you enter the search request "Nazi" in google.de, the first result is the German Historical Museum's site. With the U.S.-version google.com, search result number one is the American Nazi Party's home page. However, operators emphasize that independently of individual country's laws, search engines only remove pages from the program which fraudulently try to push their ranking to the top, and political or ethical reasons are irrelevant. Nobody can really check whether that's true. The algorithm and its criteria are kept as secret as the Coca Cola formula.
Many people slowly grow uneasy. Maybe because nobody is allowed to know according to which criteria the search engines select the answers users receive. Just a little while ago, Google was a white page, thankfully without any ads, with an innocent, colorful childlike logo, founded by two students whose company philosophy was "Do no evil". Supposedly, the former chef of the rock group Grateful Dead was cooking in their cafeteria in California. Then they went public, had Microsoft as a competitor and their company vision suddenly became slightly threatening: Organizing the world's information.
On a May evening, high above Berlin, in the attic rooms of the Heinrich BÃ¶ll Stiftung building in Berlin, some 200 people got together to discuss the world engine.
At the podium sat Stefan Keuchel, the young spokesman for Google Germany. The accusations directed at him on that evening were serious ones. "You act as if some websites don't even exist. The German Google version doesn't display some pages", said Burkhard SchrÃ¶der, editor-in-chief of the small media magazine Berliner Journalisten. "You are lying. You say that this content doesn't exist, however, it does exist. It is an attitude worthy of a dictatorship."
"Our goal is to make the lives of the users easier. We are following all existing laws", Google spokesman Keuchel replied. "If something is prohibited, we will remove such a link. But we sit between all chairs, as we say here in Germany. If we allow certain things, everybody says: Why are you displaying this? When we remove it from the index, people say we censor things."
Wolfgang Sander-Beuermann, who runs the biggest German meta search engine at the university of Hanover, which is fed by the results of thirty other search engines, was part of the discussion panel as well on that evening. He said: "I don't want to do any Google bashing, but my trust ends when I look at their monopoly. 85 percent of all German search requests go to Google. If knowledge is power, search engines are super powers. We must prevent Google from becoming the absolute super power."
"Google has no operating system which forces somebody to use Google", Keuchel hinted at Microsoft. "We are looking at the monopoly accusation in a different way. Our users use Google because they get the best results from it."
"You can't force people to use different search engines", said Katja Husen, spokeswoman for media politics in the board of the German Green party. And Sabine Frank from Freiwillige Selbstkontrolle said: "Traditional legislation doesn't get us anywhere anymore. We must regulate the content internationally."
The representative for the traditional legislative approach seemed to be a little at a loss. "Some people think it is possible to regulate the digital work in an analog manner", said the Green party spokeswomen, "but this kind of policy is doomed to fail."
At the end, company spokesman Keuchel gave an outlook into the future. He talked about the new project Google Print, which involves scanning the entire inventory of five big American and British libraries and make them available in the Internet. It was meant to be a promise, but it rather sounded like the company was planning to eat up the world.
The search engine wars have begun. Because of his worries regarding the deficiencies in spreading European culture, the president of the French National Library suggested digitizing European literature and putting it into the web. French president Jacques Chirac and German chancellor Gerhard Schröder announced to finance the project with 150 million Euros. Developers are working on the German-based search engine Yacy, which is developed under the open source principle, meaning that everybody can use it for his or her data collection and everybody can join the programming effort. In Hanover, Wolfgang Sander-Beuermann initiated the association SuMa e.V. which believes that the future lies in many regional and thematically specialized search engines whose providers are better able to filter out spam and to develop the many undiscovered spots on the Internet map.
Probably the most interesting effort to teach computers how to read is happening in Berlin. The building of the Berlin-Brandenburg Academy of Sciences covers half of the length of the Gendarmenmarkt square. It's a gray concrete block, four floors high. For 300 years, the academy has been collecting the German language, sorting, researching and conserving it. It is here that people work on a Goethe dictionary which captures Goethe's vocabulary within its historical connotations. It is here that the Prussian State Protocols are being processed and a new critical Marx/Engels edition is being worked on. The core, the "Klassiker" as they say here is "the Grimm", the biggest dictionary of word meanings in the German language which was started by the Brothers Grimm. Once you have passed the reception, there is a hallway with dark yellow linoleum floor. It is here that for 151 years, example sentences for every German word have been collected. The second of the Grimm brothers died when he had processed the letter F like fruit fly. Since then, generations of German studies specialists have continued his work.
An old-fashioned scent is in the air, maybe of cabbage soup and paper. The visitor passes quickly, takes the old-fashioned elevator to the third floor, where suddenly everything is white and recently painted. The digital dictionary of the German language is created here. It deals with the same complicated question that challenged the Brothers Grimm, namely: What do German words mean?
Five years ago, linguists and German studies specialists started to scan in 100 million words from novels, newspapers, specialized texts, operating manuals and the spoken language. Then they started to organize all of it. With a search engine written by a programmer from Moscow, a young man who did not really learn German until he started working on the project. The search engine especially needed to master German grammar, it had to be able to separate adjectives from nouns or a match a conjugated, declinated, split up or recomposed word to its root form.
Until now, the search engine has found 2.5 million different words, most of which are compound word creation like "Selbstbausarg" (Build-your-own-coffin), Apache-Kampfhubschrauberpilotin (female Apache fighter helicopter pilot), words which might have been used only one single time. As a byproduct, the young Russian's search engine remained. It is not only able to use by keywords but also knows something about the language rules. "That is the precondition for having computers understand texts", says Gerald Neumann, linguist, computer scientist and member of the dictionary team. "Sometime later." He does not want to promise too much. The futurists from the artificial intelligence research departments have said for 40 years that computers will be able to understand language soon. "I don't know when. But it will happen, because for the first time, there is a lot of money in the industry. Right now, the search engines still act as if they understood what is said in the texts. Sometime, however, they really will understand."
By a strange coincidence, on the same day, Noam Chomsky visits, the American linguist known to many as a political activist. However, he really became famous because he reformulated a basic concept of language. In the 1950s, he fundamentally changed linguistics when he developed his transformational grammar. He started with the assumption that a human being's language ability does not have to be learned from scratch, but is already wired into the human brain as an anthropological constant and that there must be common rules among grammars, some kind of genetic code for all human languages.
During the 1980s, Noam Chomsky had become a little out of fashion. There simply were to many strange individual aspects for which ever newer rules had to be created. His assumption of a depth structure of language, however, has not left the humanities since then. And now it looks like Chomsky's ideas could become interesting again, fifty years after Syntactic Structures was first published: for none other than the multibillion dollar market of automated information processing. "Wouldn't a computer be able to master this depth grammar in order to search texts in an intelligent fashion?" asks linguist Gerald Neumann. "This is because, in the long term, no ranking helps against the information flood, no matter how good, but only automatic understanding of language. And this is exactly what we're working on here."
A year ago, Apostolos Gerasoulis moved into his new house in the U.S. Since they live there, his kids have collected a laundry basket full of golf balls. This is the first thing Tomasz Imielinski is asked to have a look at when he enters his friend's home sometime in August. They are sitting at the kitchen bar. Apostolos pours light brown whiskey over ice cubes.
"At the end, I couldn't stand the university anymore", says Tomasz, who has taken a sabbatical at Rutgers. At Ask Jeeves, he calls himself "Vice President of Data Solutions". "You are working on a problem for three years, you present the solution to an audience of ten at some conference, then three of them tell you they think it's great. Then you go home and it all starts again from the beginning."
"Until you die", says Apostolos. And in order to escape this grind, they insisted on taking management positions in the company. Just like the two guys from Google did before them. "This is why I have lots of respect for Sergey and Larry", says Apostolos. "They have shown the world that scientists can lead a company, that the world is not ruled any more by wall street brokers and business types from Harvard."
On a June morning, Apostolos Gerasoulis was suddenly wide awake. It was the day investor Barry Diller had announced to visit. In America, Barry Diller is a big shot, rich and glamorous, married to society designer Diane von Furstenberg.
Even though, to some of his programmers, the name didn't mean much, thinks Apostolos. The employees had to enter it into the search engine. There, you can find the information that Barry Diller started in 1999 to acquire Internet service providers and that his company IAC survived the Internet bubble, even making a profit. Now Diller wanted a search engine. "Normally, I talk a lot", says Apostolos, "but on this day I was very quiet. It was one of those days on which you think about yourself, you let your life pass before your eyes and think: Wow, I would have never thought this could happen to me. That I could ever be in the same room with him."
Diller let Apostolos demonstrate the engines. Apostolos showed the zoom function he recently developed and which makes proposals regarding limiting or extending a search. When he entered the name "Diane von", the zoom function proposed limiting the search to Diane von Furstenberg. Diller called that incredible and said they should bring the new product to market immediately.
After that, the guest wanted to know what the future might bring. Apostolos prophesized that it would soon be possible to merge everything existing in the world into the digital sphere. Later, everything could be specifically retrieved by means of a perfect search engine. The only question is: Who is going to find the best answers in this universal library? Is it going to be him, Apostolos Gerasoulis, son of a Greek hardware store owner?
In mid July, the moment had come. Barry Diller bought Ask Jeeves for 2.3 billion dollars. Apostolos Gerasoulis smiles even a little happier than usually when he says that Diller has promised to take him to New York City soon for a big party.