Two coffee breaks and a snack run later, he felt that what he had done was enough to improve his answers substantially. He was already running 45 minutes late. It was now 2:45 p.m. He tried the new modifications on a few of the 50,000 users. The results looked better than before. He let the program generate the answers for all of the users, and avoided looking at the answers it returned. If he had looked, he wasn’t sure he would be able to resist tweaking his program further to squeeze out more improvements, all while the clock was ticking away. The moment the answers were produced, he submitted them and held his breath. By the time he leaned back in his chair, a message had appeared.
Congratulations, Task 1 sufficiently completed.
Press any key for Task 2.
Stephen surprised himself with how happy he was. Compared to the work he had done when he led his own company, the magnitude of the task just completed was miniscule. But, still, the thought of working at Ubatoo with access to all that Ubatoo had to offer (not that he was quite sure what all of it was, but he knew it was a lot), he couldn’t help but feel excited, almost giddy. But there would be time enough for all that later. Task 2 was waiting on his screen.
Data Mining Task 2: Needles & Haystacks
Out of the 100,000 users who were pre-screened to fit the profile we were looking for, 78 of them bought a car that cost over $161,000. We’re going to tell you who 40 of them are. Your job is to find the other 38.
You will have access to the emails they sent through Uba-Mail over the last year, their purchasing history when they used any of our credit cards, and the Zip code of their home address. Access to all of the searches they performed in the last year is provided.
The suggested time to work on this problem is twelve hours; we strongly suggest using outside information sources to help with this task. Whatever you find is fair game.
Twelve hours—it would be past 3:00 a.m. then. This task was big; not only because of the time allowed, but because of the data he was permitted to see: e-mails, Zip codes of home addresses, and credit cards. This was how Ubatoo made money, understanding their users, and knowing how and what to sell to them. Despite providing the tremendous number of services Ubatoo offered, it was, at its core, an advertising business, a relentless and unflinchingly efficient advertising machine.
Looking around, Stephen found half the seats empty. Maybe everyone was taking a break. He went to find food, but didn’t hear the expected sounds of all those who should have been there. “Where is everyone?” he asked the first person he encountered at the food buffet. The guy was stuck at the entrees, trying to make a decision between the salmon and the buffalo fillets.
“I don’t know,” he replied, “I think most of the others didn’t do well. I guess they weren’t given the next round of questions.”
“Oh,” Stephen replied. He wasn’t in a frame of mind to think of anything cleverer to say.
“In my group, cryptography, I heard a rumor that more than half of us were already given the boot.” Finally, he decided not to decide and heaped a serving of both salmon and buffalo on his plate. “What group are you in?”
“Data mining. Just started the second question.”
“Hope you make it. I’ve gotta run. Plenty more work to do,” the indecisive eater said as he hurried back to his seat, inhaling his meal on the way.
Stephen sat at an empty table, trying to eat slowly, trying to convince himself that fifteen minutes of rest was imperative and would surely serve him well. But to no avail. Stephen wolfed his food down in under three minutes, burned his lips and tongue inhaling a scalding hot coffee, and almost ran back to his computer.
Once again seated in front of his computer, he felt foolish for having run. The problem was still as unsolvable as before. Talk about finding a needle in a haystack: He had to find 38 people from a set of 100,000. He just stared at the screen, which still impassively displayed the second task.
Stephen shoved his keyboard back and rested his elbows on the table with his head in his hands. He didn’t bother opening his eyes. Start with what you know. Okay, first and foremost, what he knew for sure was that it was solvable. The information was accessible, and it was right there at his fingertips. Zip codes, e-mails, spending habits. What does a Zip code reveal? It could reveal a lot: how wealthy a person is, what type of house they owned. All of that information was publically available on the Web—the Census Bureau collected that much and more. He also knew he could eliminate almost anyone who shopped at GreeneSmart too often, come to think of it. He hadn’t seen a $161,000 dollar car in the parking lot in quite a while. But what other patterns were there? He could eliminate anyone who regularly did their own errands or shopped at the dollar store. Okay, this was all available from their credit card usage information, to which he had access.
What else is an indication that someone will buy an expensive car? Their jobs? Maybe. That correlates well with Zip code, too. What else? What would convince him to buy a car that expensive, he wondered. Nothing, really. But his friends might try to convince him. Which friends? Those friends who had purchased some crazy car themselves? Yes. If he was the type of person who bought a car for $161,000, he might try to convince his friends to buy one, too. At the very least he’d tell his friends about it. Some of that must be in e-mail trails. What about pictures of the car? That would probably help, too. And if you were about to be convinced to buy a car, wouldn’t you search the Web and research it? He certainly would.
So, let’s start the groundwork. Find out which cars cost over $161,000. We know forty people who bought them. Who were their friends? Read their e-mails. See who they told about it. Which ones of them also searched for these expensive cars? We have the names of the cars, and now we have the potential set of friends who bought them. We know where they lived and approximately how much their houses were worth, too—just by looking at the Census Bureau’s stats and a few real estate web sites. Could it be so easy to put this together?
Thinking this through, Stephen was as still as humanly possible, except for the small rapid eye motions beneath his closed lids. The lack of apparent motion concealed the swarm of activity in his mind. His thoughts scurried in a dizzying game of tag, charging through the minefield of false leads and the countless pieces of data requiring analysis to expose the missing thirty-eight people. Eleven hours remained.
-CONTACT-
March, 2009.
Half of Stephen’s shift had already ended by the time he woke up at 1 in the afternoon. Thirty minutes later, he was in the children’s department at GreeneSmart, on his way to find Molly. She was the only person he wanted to see.
“Well?” Molly asked anxiously, as she peered from behind a clothing rack. “How was—”
“It was incredible. I haven’t had to think that hard in years. The people there were so smart . . . 600 of us were there . . . just working, hacking away on these ridiculously difficult problems. It was unbelievable. We were working until 4:00 this morning.”
Even Stephen was taken aback by how fast and excitedly he was talking now, and he wasn’t sure how to stop. “. . . and the data they had, well you understand how hard it is to get data, and they had it all laid out for us, and they just let us play with it, dissect it, and . . .”
By this time, Molly had taken his arm and guided him to a slightly less crowded aisle, where fewer people were watching them. Then, she finally inserted her fourth word into the conversation, “Well?”
Stephen stopped in mid-sentence, “Oh, I made it. They made four offers for my group this year, and I got one of them.”
Molly was already giving him an enormous hug. “Congratulations! I knew it. Let’s go out to celebrate. It’s on me.” She gave him another hug. “But you might want to change first—you’re a mess,” Molly said, happily. She grabbed Stephen’s hand and started walking back toward the exit doors as if encouraging him to go home and change that instant.
“Sorry, I didn’t have time to get ready. I just wanted to come and tell you
about it.”
“I’m really glad you did. That’s all I’ve been thinking about. When did you finally get back to your apartment?” she asked as they were walking.
“Around 6 this morning, but we were at Ubatoo until almost 5. It was a complete disaster . . . baseball with Coke cans in the hallways, others running around the desks playing laser tag, being totally obnoxious. Then, the PlayStations and Xboxes came out. They projected them on this giant four-story screen in the room we were in. Okay, that was actually kind of cool.”
“Sounds great.” Sarcasm. Eye roll. “Out of control, sugar-high kids. Delightful, Stephen.”
By this time, she had successfully led him to the front doors. “Well, you made it. That’s huge. Tell me more about it tonight, okay? What time do you want to go out to celebrate?”
“I think I’m going to have to take a rain check,” Stephen hesitatingly replied. “They’re having a party tonight for the new data-mining interns. I think I’m expected to be there.”
A look of disappointment crossed Molly’s face for an instant, but just as quickly disappeared. “Well, the offer stands whenever you want,” she said as she gave him a quick kiss and started walking away. “By the way, congratulations again,” she called back to him.
“Hey, Molly!” he yelled out, loud enough to startle the kids and moms walking by. She stopped and waited. “You were right! Thanks!” With that she disappeared behind an aisle full of office supplies.
That night, the four data-mining interns, distantly familiar from the previous night, were looking at each other uncomfortably across a formally set, candle-lit table to which they had been directed. Although they had no notion of what to expect, it certainly hadn’t been this.
“Anybody know what this place is?” asked Stephen, the first to speak. His voice was barely audible over the noise of all the other partygoers, some Ubatoo employees, others from Ubatoo’s large advertising clients. Stephen pointed to the two frescoed palatial domes in the four-story ceiling above him and shrugged. The surroundings married the imposing architecture of a gothic church with the gaudiness of newly minted millions.
The only woman in the group of four, Aarti, answered, “This is Xiao’s house.” She spoke with the speed and deliberateness of a southern drawl, yet with the unexpected sophistication of a British accent.
All eyes were upon Aarti and the oddity who happened to be sitting to her left, Kohan. It was a peculiar sight, seeing the two side by side. Both were Indian, but that was where any similarities ended. Kohan was a short rotund man with a prominent thick black moustache that he was constantly pressing and smoothing down with his fingers. If his hand wasn’t at his mustache, he was typing something furiously into his BlackBerry, which he had already done numerous times since they sat down. His plaid shirt with cowboy-styled piping over both pockets, complete with bolo tie, oversized belt buckle, and brown leather riding boots, provided a perplexing thought experiment as to what calamity could have possibly produced this modern Indian Cowboy.
Then there was Aarti. She could best be described as Audrey Hepburn with a tan. Everything about her, from her tiny body to the way she was dressed (with her fitted pencil skirt and demure dove grey sweater set) to her hair, set in a neat twist at the nape of her neck, put her in a class different—above—everybody else at the table.
“Xiao had to buy and tear down six houses to build this place. This—this is what Ubatoo is all about,” Kohan replied reverentially. They looked around at the marble expanses, the several stories high glass walls, and the paintings that they now suspected were originals. “The question I have is: Why are we here?” asked Kohan. “Maybe we’re the entertainment,” he said again, answering his own question with a smile.
William, so far silent, took his turn in adding to the conversation, “Our being here is likely an ill-conceived scheme some marketing person came up with at the last minute to show how big the data-mining team is. I wouldn’t let it go to your head. I can’t imagine they really care if we’re here or not.”
“You’ve become more of a curmudgeon than I remember,” Aarti declared, a healthy dose of disdain in her voice.
“No more so than usual,” William replied irritably.
“Don’t mind him,” Aarti explained to the others. “I’ve known William for years. His humor is on the dry side.”
“Champagne?” A voice called out to them. A man dressed in a uniform of red and gold with a giant peacock feather looming high above an exaggerated turban on his head was holding a tray out to them with crystal flutes filled with champagne. Stephen grabbed one immediately, as did everyone else in turn, except Kohan, who didn’t hesitate in grabbing two.
“Gentlemen, and gentlewoman, of course, I propose a toast,” Kohan said. “To arriving, and arriving in style.”
Taking Kohan’s lead, Stephen tried to change topics. “That was tough last night. I’d heard the entrance bar was high, but that was beyond anything I imagined. I struggled for an hour before I had an insight into the car buying problem.”
“Last year, when I was an intern, we did that kind of work all the time. This one was almost a textbook case,” Kohan replied. Then he quickly added, “But if you hadn’t done this before, I can imagine it being pretty tough, especially in a time crunch.”
“The first question was trivial, if you ask me,” William said. “I finished that in half an hour. The second one, I wound up finding the group of thirty-eight in two different ways just to double-check my answers. No issues here.”
Stephen thought that possibly “curmudgeon” wasn’t an appropriate description for William—the little runt of a man. Not to stereotype, but it was hard not to picture William as the maladjusted, snippy, peevish, repulsive part of every stereotypical computer scientist. Maybe “jackass” was a more fitting description.
“Everybody having a good time, I hope?” a voice called to them. Before anyone felt compelled to try a third attempt at conversation, Atiq had thankfully found them. He was dressed sharply in a black tuxedo with black tie. “I’m glad to see you all could make it.” He looked around the table happily. “I wanted to tell you what’s in store for the four of you tonight. Xiao is going to introduce you as Ubatoo’s next generation of knights in shining armor. He’ll probably tell everyone that, during your internships, you’ll work with them to help them learn more about their customers. It’s all part of the ‘Touchpoints’ program—our data-mining initiative you’ll be working on.”
Atiq spied worried expressions on at least three of the four faces—excepting William, who was confident he would do a good job.
Atiq continued, “Don’t worry about it. Normally, we have a team who does the business side of things. Sometimes, though, it’s good for the advertisers to talk directly to the eggheads in the operation. It makes them feel good. All you have to do is be enthusiastic; they know you haven’t even started yet. I’m sure you’ll dazzle them. Trust me, it will go well.” He looked around for a moment and smiled at Kohan, “By the way, nice shirt, cowboy.”
Stephen couldn’t recall a single employee he’d ever hired at SteelXchange who he’d let within ten feet of a customer without extensive training. Maybe this was another indication of the type of talent Ubatoo managed to attract. Or, maybe, interns could do very little that would actually hurt an enormous company like Ubatoo. The latter thought comforted him a great deal.
Atiq reached into his pockets and pulled out four decks of business cards, one for each of them with their names proudly displayed next to the official Ubatoo logo, and a title of “Software Intern—Touchpoints Team.” E-mail addresses and phone numbers were already assigned and waiting for them to start.
“By the way, why are all of you sitting by yourselves? Meet some people. They’re quite friendly, I promise.” Atiq turned around and motioned for someone to join them. “I’d like to introduce you to Miss Erion,” he said. “It’s thanks to her company, Aston Martin, that we had the second question in the internship contest. We work
ed on that very problem for Miss Erion last fall.” Introductions being made, Atiq hastily departed to find the next set of people to meet and greet—another table of helpless guests needed to be rescued.
“What were you doing before Ubatoo?” the man asked Stephen. The man, who had introduced himself and his company a few minutes earlier, was still completely anonymous to Stephen. He was the fourth to talk to Stephen since dessert had ended less than thirty minutes ago, and nobody’s name had stuck. Stephen replied mechanically about his startup, SteelXchange. Although he had muddled through the answer the first time he was asked the question that night, he knew better than to talk about GreeneSmart again. It had led to numerous increasingly derogatory questions that he didn’t care to face a second time.
“You mind if I have someone from my team call you sometime? I keep hearing about the great returns you guys get for all the other advertisers.” The man spoke with a slur, the sour telltale hints of too much alcohol invading the air between them with his every breath.
“Of course,” Stephen replied. “Here’s my card. I’ll do whatever I can.”
“Listen,” the man continued in a more hushed voice, putting his hand clumsily on Stephen’s shoulder, “when your internship is done, we could always use smart entrepreneurs on my team.”
Stephen had heard that job offers came easily for anyone associated with Ubatoo, especially for interns. Still, getting offers before your first day of work was a bit unexpected. If the grapevine was accurate, the most coveted offer, the one from Ubatoo that would make an intern a full-time member of the team, was rare.
The Silicon Jungle Page 5
