The Silicon Jungle

by Shumeet Baluja

work address

  addresses explicitly provided

  addresses automatically extracted from correspondences

  age inferred

  age explicitly given

  birthday

  maiden name

  nickname

  viewing history – videos watched

  viewing history – online TV watched

  length of online activity

  time between online activity

  location history given

  phone number

  location history inferred

  calendar entries

  online activity timings

  personal photos

  faces recognized in photos

  objects recognized in photos

  travel destinations recognized in photos

  clothing recognized in photos

  search terms entered

  ads clicked

  ads shown but not clicked

  web sites visited

  web sites receiving a click

  images viewed

  images clicked

  image magnification

  image linger time

  sexual orientation inferred

  sexual orientation explicitly given

  mobile phone call logs

  purchase transaction records

  billing records

  bill payment dates

  profile updates

  profile update frequency

  personal documents

  stock portfolios

  credit checks

  credit card numbers

  credit card purchases

  credit card spending trends inferred

  credit card balances

  travel information given

  travel information inferred

  personal files uploaded

  spreadsheets uploaded

  music preferences given

  music preferences inferred

  medical records

  medical conditions inferred

  preferences implied

  preferences stated

  Social Security numbers

  instant messages

  instant message statistics

  searches on maps

  searches on mobile phones

  typing speed

  hours spent online

  sms messages sent

  sms messages received

  friends and contacts stated

  friends and contacts inferred

  social network stated

  social network inferred

  The list went on for six pages and had 220 items in it. The document ended with the statement that it was also reason for termination if, for non-company business, any individual’s information was identified, unencrypted, un-anonymized, or even examined for personal use, monetary or otherwise.

  The strongly worded document was intended to instill respect for the privacy of all of Ubatoo’s users. But, at least for the four sitting at the table, it would have been difficult to construct a better combination of words to get them more titillated than ever before.

  “Six full pages!” Kohan exclaimed. “I bet there’s more data at Ubatoo than has ever been collected by any company. Probably more information on people is on their, I mean our, servers than has ever been collected in all of recorded history.”

  No one said anything. “Judging from the lack of arguments, I think we all agree. That’s a first for us,” Aarti said, smiling.

  The first thing Stephen wanted to do, the minute he had access, was to find out what they knew about him. Though he did wonder if that was within the legal bounds of the documents he just signed. He didn’t think it was, and he was right.

  Each time someone used Ubatoo’s services for sending an e-mail or typing an instant message, a copy was stored in their massive databases. Every search ever conducted, whether at midnight in the privacy of your home or during business hours for your company, was the same for Ubatoo; each was logged and queued for analysis. Update your profile for your friends to see, and instantaneously seven copies of your page were updated across all of Ubatoo’s servers, no matter where they were located in the world. All the presentations and the files you stored with Ubatoo for safekeeping and for sharing with colleagues, mothers’ groups, and company boards were redundantly backed up in Ubatoo’s “cloud”—vast farms of servers located in dozens of sites on six of the seven continents. The exact locations of these farms were known only to a few people within Ubatoo.

  Where were all the places that your file was located? Nobody knew; only a sophisticated set of algorithms determined where the content was stored and where it was backed up. The algorithms automatically adapted to usage patterns, volume of inflows, and size of requests, just to determine where best to store your book report, your MP3 files, your stock portfolio, your e-mails, and your electronic greeting cards. The benefit to you was that even if all of Europe suddenly was hit by a meteor, you would still have multiple copies of your e-mails and pictures of your vacation ready for perusal without interruption.

  And when these servers weren’t busy writing and storing your data, what did they do? Each set of servers crunched the data they had amassed, twenty-four hours a day, seven days a week, 365 days a year. No break required. Click something on Ubatoo’s site, find a search result you wanted to read more about, an image you thought looked good? Immediately those clicks triggered dozens of machines to spring into action, ensuring those clicks were logged and saved for analysis. Even more immediate action was taken when someone used an Ubatoo phone or Ubatoo’s virtual or plastic credit cards. These were high-value pieces of information that revealed much more personal and interesting insights about you, and so were queued to be processed immediately.

  What the interns suspected was accurate: Each piece of this massive information repository amounted to more information about an individual than had ever been collected before, in all of recorded history, integrated through time. Although Ubatoo may not know the color of your eyes (unless you decided to upload a picture of yourself to their photo sharing services), enough data existed on their servers to tell a detailed story of you, down to the most minute details of your life. Everything was there, from your buying habits, to the e-mails you sent that you wouldn’t want your boss to know about, to the pictures you searched for behind the privacy of closed doors. And why was this information collected? The simple answer was so Ubatoo could show you a better advertisement.

  Create a profile of every user and utilize those profiles to personalize the temptations they see—flash an ad before their eyes for something Ubatoo knew they would crave before they did. Everything Ubatoo did, and every service it provided, seemingly for free, actually came at a miniscule price: your time, your attention, and hopefully, your click. An advertisement was shown to you. If you clicked the ad, the advertiser paid Ubatoo. Though almost nobody clicked regularly, almost everyone clicked sometimes. In the end, it was a simple numbers game. Make services that people used so they kept coming back, and each time they did, show an ad. Eventually, something would be enticing enough to make you click.

  Adamantly against clicking? Not going to be swayed? That’s okay, too. The profile built on you helped to understand others. Plenty of people like you existed, people who searched on the same things, who spent the same amount of money per week, and maybe even who lived in a house close to yours, and these people were more amenable to advertisements. So use the services all you want, your data was as good as your neighbors, and both sets of data were equally welcome.

  The extremely grounded and ruthlessly capitalistic mission of better advertising was well understood among everyone who worked at Ubatoo. It stood in stark contrast to the idealistic and infamously liberal demographic of the “talent” employed there. This did not go unnoticed by management. For most of the workers at Ubatoo, the mechanisms behind Ubatoo’s capital successes were placed in the same category as knowing that the sun rose in the morning: It mattered—they wouldn’t k
now what to do without it—but it wasn’t something that needed attention. For the technologist who worked on the “front-lines,” there was the everyday battle of keeping the almost 3,650,000 servers located around the world up and running. For the hundreds of engineers working on their e-mail service, little was on their minds except meeting the growing requirements of a user base that was expected to triple in size by the end of the year. For the several thousand Ph.D.s working to make Internet searches more efficient, the thought of money and advertising never crossed their minds; the rapid growth of the Internet and all the information being produced was enough to keep their intellect entirely occupied.

  For almost everyone at Ubatoo, the pace was so frantic that the only way to keep up was to keep their heads down. Immerse yourself in whatever project you were working on; everything else was taken care of for you. The only people who worried about finances were the “money people,” who were intentionally housed in a building on Ubatoo’s grounds separate from the scientists and engineers. The influence of the money people on the technologists was small; they were never allowed to make decisions regarding products or the activities of the technology talent. Instead, the sole job of a tiny number of selected individuals was to bridge the gap between the technologists and the money people. These selected individuals had two offices, one with the technologists and one on the other side of campus with the money minders. These individuals shielded both sides of the company from each other.

  It was because of this carefully constructed shield that the four data-mining interns could remain blissfully enamored of the company despite, or in their case, because of, the papers that they had just signed. The thought of using all this data for producing money wasn’t a connection they had made. What they knew, instead, was that they were about to be offered the keys to a kingdom of pure raw information. Few people, outside of Atiq’s group, ever considered the number of actions, thoughts, and desires of users stored in Ubatoo’s cloud, waiting to be mined.

  And absolutely nobody, other than Atiq, ever grasped the enormity of what it meant to have profiled almost every user on the Internet. Almost everyone who had touched a computer had used at least one of Ubatoo’s services at some point, whether it was through one of their partners or through Ubatoo directly. For each user, a profile had been constructed, waiting to be filled in with countless tiny bits from their lives, significant or otherwise, be it done over an hour or a year. All that the interns knew so far was based on the document they signed, which meant all they really knew for sure was that there was a lot of data. They would soon realize the magnitude of such an egregious understatement.

  -EUPHORIA AND DIET PILLS-

  June 1, 2009.

  Two-wheeled razor scooter, flames painted on the handles, driven by an unidentified emaciated Indian guy with really big hair—heading straight for the table. He stopped inches from Kohan’s boots and smiled as he looked over the cowboy. Without dismounting the scooter, he said, “I’m Jaan Ramamurthi—your sponsor. Follow me.”

  If there was such a thing as “royalty” in the data-mining community, Jaan Ramamurthi certainly qualified for that distinction. Although there may be fewer people in the world who recognized Jaan compared to the other more widely known royals found in castles and the tabloids, the respect his name evoked at this table certainly rivaled the respect any royal hoped to receive. Of the four, only Kohan knew what he looked like. He had watched Jaan accept the prestigious Breakthrough Award at the invitation-only banquet held in his honor at the Pattern Recognition and Machine Intelligence Conference last summer.

  As they left the building and tried to keep up with Jaan and his scooter, they were surrounded by idyllic greenery, the obligatory gently gurgling fountains, and pristine building façades. All of this was in stark contrast to the scene that soon emerged. The stern concrete buildings to which they were headed were a drab mix of dark tinted brown glass and the just-slightly-less-brown concrete. From the outside, the buildings were as functional and austere as could be designed. Every hundred feet or so, a sign was posted next to the walkway: “This area is for employees only. Absolutely no guests permitted.”

  Jaan finally dismounted his scooter outside Building 11, in the mini parking spaces especially designated for scooters. A little receptionist, barely visible behind a high brightly colored desk, pressed a button, and a large door behind her left shoulder silently opened. “Good morning, Jaan,” she called out loudly, the first sound they heard other than Kohan’s boots on the shiny concrete floors. Jaan passed her with barely a glance, and a quiet mumbled thanks. The four followed close behind.

  The labyrinth of corridors they had expected to find was instead a large open room with a mix of brightly colored low cubicles in the center, and small glass offices around the outer perimeter. Not only did it allow one to see from one end of the enormous room to the other, but it also allowed natural light to be seen from everywhere. A soft constant humming permeated the room—the sound of hundreds of computers running, waiting for something to do.

  The offices were nothing special to look at, but they did house some of the more senior people at Ubatoo or those who had earned their place by performing a superhuman engineering feat or product release. Each office held two or more whiteboards filled with countless arrows and boxes and barely legible formulas, all liberally sprinkled with giant “DO NOT ERASE—¡No Borres!” magnets.

  Building 11 was the home of the research group. The nameplates identifying the occupants of the offices were the latest source of wonder for the interns. The nameplates, subtly lit with solar-powered LEDs, held the names of the professors who wrote the textbooks they had used in graduate school, the lecturers from whom they had taken their courses, and the numerous classmates and school alumni who they had heard about or had vaguely known. They had all managed to find their way to Ubatoo. No other institution, in academics or in industry, had been able to amass such a stellar cast of computer scientists in its halls.

  Most of the scientists were amicably chatting away or staring out their windows with glazed expressions. If work was in progress, it certainly wasn’t apparent to the casual observer. Like any research lab, innovation happened in fits and starts. Long periods of frenzied work and creativity were followed by stretches of thought and play.

  When Jaan stopped suddenly, the four interns who had been walking single file behind him, as if taking a cue from any Saturday morning cartoon, abruptly crashed into each other in their attempt not to run into Jaan. Jaan did his best to ignore them, and instead pointed to a group of about seventy desks in the middle of the room.

  “Find good spots to sit in. Most of the desks will be taken by day’s end. Interns from the other groups will also be sitting here. As for computers, I’m sure someone from IT will be around later to help you get set up. You don’t have to worry about any of that—let someone else handle it,” Jaan directed.

  Jaan gave them a few moments to look around and scope out potential spots before continuing. “Normally, the first set of intern orientation meetings would take place at 11 o’clock. I’d like you to skip those. Instead, let’s have our first meeting.” He waited to see if anyone objected, but no one said a word.

  “So, how about taking thirty minutes to drop off your stuff and find your way around? The snack rooms are back towards the lobby, as are the all-important coffee baristas. When you’re done with that, meet me in the Huckleberry Finn conference room at 10:30. See you in a few minutes.” Jaan then left them, and walked about a dozen feet to his own office.

  Without knowing what else to do, the four found seats close to each other and near Jaan’s office. Next, once the four confirmed what they had heard was true—all the perks and food and drinks on the grounds were free—the first and only stop during their thirty minutes was the closest coffee bar they could find. With double cappuccinos and extra-large gourmet coffees in hand, along with an enormous number of cookies, biscotti, and candy in their pockets, the four interns made their way
at 10:25 to the Huckleberry Finn conference room. All the conference rooms were named after famous figures in children’s literature.

  Jaan was already seated and waiting. “Excellent. Good to see you found the conference room.” They unloaded their pockets onto the table. “Glad to see you’ve already begun looting the place. It wouldn’t be intern season if you hadn’t.”

  “Let’s get down to business, shall we? Atiq is in a bit of hot water for only hiring four of you for the Touchpoints project, so he’s asked me to step in for a few weeks while he goes on a recruiting tour again. I’m going to make sure you get up to speed quickly. The best way to do this, I imagine, is just to let you jump in and do something constructive.” No comments arose, so he continued.

  “I was thinking I’d just give you the project I’m working on. It’s a marketing study, and not quite as sexy as selling sports cars like you did last time. I need you to find buyers for diet pills. Why, you may ask, do we care about diet pills? Well, it turns out . . .”

  Diet pills. The long anticipation Stephen had felt in coming to Ubatoo was culminating on his first day with selling diet pills. It was, without overstating the obvious, an utter letdown. This is not the way he had anticipated spending his summer. He vaguely heard Jaan go on, rather unconvincingly, about the market capitalization of diet pill manufacturers, how much they spent on Ubatoo’s sites per minute, the importance of this market, and so on.

  Jaan finished his discourse with a question to all of them. “Now, it’s your turn to talk. How would you start finding the best buyers for diet pills?”

  Stephen tentatively raised his hand to offer a suggestion, but William started spewing half-baked thoughts without a moment’s hesitation. “I think we can easily tackle this using the formulation in the paper by Redmon and Soffti that was presented at this year’s Knowledge Discovery and Data Mining conference,” he said with authority. “They showed the effectiveness—”

  Jaan didn’t let William finish his sentence. “You’re talking about the paper in which they analyzed the buying habits of several hundred of their students over a period of two weeks, right?”