Gu had been thinking about problems he could solve through technology. That was beginning to become the standard refrain among aspiring entrepreneurs: find a problem to fix. Lately, founders had become so overzealous in this endeavour that problems were created that never existed, or problems were over solved to the detriment of practicality, such as iPad menus at airports, which took live servers to explain to frustrated users. It took Gu six months of working in New York to develop some real problem that wasn’t being attacked from all sides by 20-and 30-somethings salivating to stumble upon the next TaskRabbit – a company that lets you find strangers who will do small tasks via an app. But then Paul realised the biggest unsolved problem was all too obvious: he knew too many contemporaries who were broke.
Some of his friends weren’t doing so well, either. Gu didn’t have a credit history, and he had no home to borrow against. He wasn’t sure what to do. What if people could loan him money directly? Maybe they could do it online or through an app. Maybe he didn’t even have to go through a bank.
He had also met his future cofounders: two Google employees who were 20 years older than he was and thought this idea of his could become a real business. They had been thinking about the same issues young people had, but they already had jobs — good ones. One of them, David Girouard, was president of Google Enterprise, a line of business products such as Gmail for businesses, which was rebranded in 2014 as Google for Work. They were established in Silicon Valley. Gu considered him the perfect partner. Girouard had the experience, and Paul had the algorithm that would use metrics such as a GPA, the person’s career goals, and past internships to predict future salaries. They would join forces to start a new company called Upstart and would fund mostly young people who couldn’t get a line of credit or a loan. Seven other student entrepreneurs wanted to participate, and the beta-testing team was formed.
In spring 2012 he and his new cofounders raised money within a few weeks to get started, thanks to their connections through the Thiel Fellowship. The Founders Fund invested too. They worked in office space reserved for incubating companies in Google Ventures and then Kleiner Perkins — right on Sand Hill Road. It was like they’d arrived. But they hadn’t, really; they were merely where it was all happening. Upstart had an algorithm, but it needed to find a seamless way to make the company work that would benefit both investors and those seeking loans. “We sort of had this general concept of a problem in mind, but the solution was quite different than what we are building today,” Paul said.
Meanwhile, his job was to assess risk in the same way that East Cost businesses and industries did. His whole company used success metrics, such as GPA, to predict future success. At Upstart, Gu focused on specific ways to do that. “It’s not whether you can pay [back the loan]”, he maintained. “It’s a question of how important you see your obligation.” He realised that people who were diligent in school and studied conscientiously tended to honor their commitments. It was a way to judge people through data. In a place where social skills were poor, to put it mildly, algorithms served as a more reliable predictor.
Gu went so far as to say the algorithm even judged a person’s character. Indicators pointed to warnings that someone might not necessarily be reliable, such as if he or she had used prepaid wireless numbers — a possible warning sign that the person didn’t have a steady paycheck.
So now Silicon Valley was even trying to crunch data and numbers to determine character. Perhaps this data could predict a person’s morality, responsibility, and reliability. Maybe cloud-based personnel software could replace humans perception, and help human resources departments predict when employees will quit, how they will perform, and how long someone will keep a job. It characterised managers as “rainmakers” or “terminators.” Analyzing employees through key words, such as a Google search, was becoming more and more common as even something as ephemeral as character was becoming digitized.
In one sense, the digitisation of character was a way to remain politically correct while still sticking to the facts. The data didn't lie, right? Instead of offending anyone with subjective descriptions, digital character judgements were objective. They had predicted crime, after all, with systems such as CompStat in New York, a computer statistics management tool introduced into the NYPD in the mid-1990s. No one had to say, "That's a seedy neighbourhood composed of this race or that race." The data did it for them.