March 28, 2014 by Speakers' Spotlight
Greg Lindsay on Relationship Science: Harnessing Big Data for Power Networking
Greg Lindsay reports from the front lines of globalization. Described as an “intellectually stimulating and thought-provoking, educational and entertaining, absolutely fascinating and superbly presented” speaker, he shares fascinating information about our globally interconnected future, in this article Greg examines a new platform for networking:
Call it a more sophisticated LinkedIn. It might just be the magic bullet for business development.
Maybe you know Neal Goldman. Or maybe you know someone who does. A minor celebrity in both Davos and Big Data circles, Goldman sold his first company for $225 million, back when less than a billion dollars was actually worth something. Tonight, I am meeting him for the first time, sharing a train ride to Philadelphia, where I’ll get to hear him pitch his latest company, Relationship Science.
While waiting in the maelstrom of NYC’s Penn Station, I run through what I’ve learned about him: where he went to school, where his sister-in-law works, and how much cash he donated to Rudy Giuliani’s presidential campaign ($6,900). I also go over a mental list of people in his circle–including, crucially, the ones we have in common.
I’ve learned all of this because before I met Neal Goldman, I stalked him, using his own software. So when I finally pick him out of the swarm, I’m able to drop the name of a mutual friend: “Parag Khanna sends his regards,” I say, referring to the foreign-policy wonk and author. Goldman, wearing the tech exec’s standard-issue scruffy beard and horn rims, rocks back on his heels, eyes vaguely searching, trying to put the pieces together. And the fact is, I don’t have a clue whether my old friend Khanna and Goldman are drinking buddies or mortal enemies. I’m about to find out.
Relationship Science–or RelSci–is an online platform built with profiles drawn from the 1 percent. It’s not some free-for-all social network full of selfies and botspam, like Facebook or Twitter. Nor is it LinkedIn, whose metastatic expansion threatens the integrity of the entire system. And unlike net-worth-obsessed virtual clubhouses such as A Small World, RelSci isn’t for swapping houses in Gstaad or getting tips on a fabulous butler in Ibiza. In fact, the point of RelSci has nothing to do with expanding your circle of friends. The point is to use the people you do know to find a pathway to the rich and powerful ones you don’t.
Goldman likes to call RelSci “the Death Star of business development.” The company’s software gives users a simple, graphical view of how members of the Establishment are connected and how those people may be connected to you. The enterprising Machiavellian can then use that map to plot a route to new relationships. And new customers.
RelSci says it already has three million-plus people in its database, but that number is less important than expanding the number and quality of links among them. “We’ve chosen not to be user-generated,” Goldman insists en route to Philly. “Users can fluff anything.” That’s why you don’t join RelSci, you can’t quit, and there are no endorsements. RelSci–not you–decides who’s in and who’s out. If you’re one of the thought leaders, decision makers, or entrepreneurs who matter in RelSci’s world, you might have a profile on its database and never know it; Larry Page, Marc Andreessen, Jeff Bezos, and Tony Hsieh are there whether they want to be or not.
And odds are, you aren’t there: Pay your individual $3,000 annual access fee and you’ll likely find you don’t even rate a spot on RelSci’s list. (For-profit enterprise fees begin at $9,000.) “We’re not a social network,” Goldman explains. “We’ve built a matrix of how the world really works.”
RelSci uses the people you know to find a pathway to the powerful people you don’t.
Already, universities and non-profits are using the system to identify new donors; bankers, lawyers, and head-hunters prefer it for burrowing into prospective clients; and investors scour it for insiders who can satisfy due diligence ahead of deals. The model is apparently so enticing that Goldman won hundreds of clients and, the company says, nearly eight figures’ worth of business in 2013, which was RelSci’s first year since coming out of stealth mode. Equally impressive, he raised approximately $90 million from such old school moguls as Henry Kravis, Ron Perelman, and Home Depot co-founder Ken Langone–men whose deal-making prowess is largely a function of their contact lists.
Much of what RelSci is selling is the accuracy and objectivity of its data. RelSci isn’t built from the kinds of self-reported delusions or exaggerations that lie at the core of social nets. Instead, Goldman and his team of several hundred engineers, editors, and data scientists (many based in India) have quietly spent years constructing their network using only publicly verifiable information. This not only requires constantly scraping the Web for updates but also building rich profiles from tens of thousands of databases, ranging from SEC filings to paparazzi photos to tax records. These pieces have in turn been joined to link everyone through past and present employers, board memberships, investments, donations, politics, and even siblings, children, and spouses.
Relying on news reports and databases–rather than the users themselves–has obvious strengths and weaknesses. At its best, RelSci is a fine-grained, fact-checked compendium of the world’s most influential people and the web of affiliations that binds them together. In practice, however, relying on editors and public documents leaves the database riddled with holes, especially in areas outside its core of finance.
RelSci may have the full measure of a man like Ken Langone, but luminaries in Hollywood or Manhattan media circles may have skeletal profiles, if they exist at all. Goldman counters that the database is gaining resolution all the time. He’s betting he can digitally reconstruct the interpersonal links, upload them to the cloud, then leverage Big Data’s ability to connect dots invisible to the naked eye. He sees his software as a tool for mining the value buried in our latent connections. If Big Data is the new oil, Goldman likens RelSci to fracking: “We knew it was there, but we couldn’t tap it. Now we have the technology to bring it to the surface.”
Using RelSci is a bit like using the New York City subway map. When you’ve identified your destination–in my case, one Neal D. Goldman–the software produces a color-coded schematic of the various routes to get there. Each station represents a person, and each line between people defines their relationship. For Goldman, RelSci presents me with 300 choices, all within three degrees of separation. Its proprietary algorithms sort each link into strong, medium, and weak ties; mousing over a name or track reveals the details of our connection.
There is no contact information on RelSci. If I have to ask how to find my friend Khanna to get an introduction to Goldman, then I don’t really know Khanna. This last point is crucial: RelSci doesn’t open any doors. It doesn’t make introductions. It makes clear the connections you might exploit, but leaves the exploitation up to you–which explains why even Goldman’s backers are equally horrified and gleeful about his product. “Most of them told me, ‘I hate it! I love it! I have to have it!’ ” he says.
“What he’s selling here is that we’re all interlinked,” says Langone, who sits on the board. “And damn it, it’s scary! What used to take two or three days of calling around now takes 15 minutes. And all of the information is totally public, so if you have a problem with how I got to you, don’t bitch at me; bitch to the SEC.”
As with any network, RelSci’s utility rises with the breadth and depth of one’s connections. Machers with a vast global web, such as Langone, are much more likely to find a short, strong path to their prey than the poser who lofts a Hail Mary on the basis of a shared alma mater or long-ago internship. The Chinese call it guanxi; we might call it juice (or Klout). And as quickly becomes clear, I have very little of it.
One exec challenged Goldman to get her kid into a private school. With one click, he delivered.
To put the RelSci software through its paces, I decide to use it to find Reid Hoffman, the semi-reclusive LinkedIn co-founder and Greylock partner. The software charts an easy path for me through Mark Gilbreath, the amiable CEO of an office-space startup called LiquidSpace in which Hoffman is an investor. Gilbreath is happy to help, but his emailed introduction is promptly batted away. It doesn’t matter that Hoffman trusts Gilbreath; there’s still no upside in talking to me. So I try to track down Y Combinator’s Paul Graham. Again, RelSci illuminates an easy line, this time through PopTech impresario Andrew Zolli. Alternately amused and aghast by my call, Zolli confirms he knows Graham, but declines to help anyway, as passing me along would draw down his own stock of juice.
“It’s not about who I can introduce you to,” he says, “but who you genuinely know. I think [YouTube comedian] Ze Frank said it best: I think about someone who has 2,000 friends on the Internet the same way I think about people who have 2,000 sexual partners.”
Goldman has come to Philadelphia this morning to address his peers in the information business, those whose databases form the backbone of RelSci. Dressed in his backwoods preppy uniform of a down vest over khakis and a rumpled shirt, he starts by wishing for “a tool that would exponentially increase my ability to sell.” Then he lists the challenges facing professional networkers like himself: “Who should I be talking to? How do I create serendipity?”
At breakfast, our conversation had quickly veered toward serendipity, a current Silicon Valley obsession. And it turned out that Goldman had profited from it once before. He had been a junior investment analyst at Lehman Brothers, killing nights filling spreadsheets, when he decided there must be a better way. In 1998, at age 27, he quit to start Capital IQ, peddling what amounted to an analyst-in-a-box to his former employers. A few years later, he sold the entire company for $225 million to Standard & Poor’s after a routine sales call.
But when it comes to meeting people, Goldman believes that what we call serendipity is merely the random expression of dormant connections all around us. To make the most of those possibilities, networking must become systematized and ultimately productized. “It’s all about the combinatorial possibilities emerging from these unknown but already existing relationships,” says Josh Wolfe, a managing partner at Lux Capital and personal investor in RelSci who’s also a complexity-science maven. “I guarantee you and I probably have five things in common we’re really passionate about.” (One of them, it turns out, is our mutual friend Parag Khanna.) “If I have a tool that helps me discover those things…that’s immensely valuable.”
Talk to Goldman long enough, and it’s clear that for all the hype around “social business,” we’ve barely begun to visualize, much less methodically exploit, the social networks everywhere around us. In one survey, nine in 10 of the executives polled agreed that the strength of customer relationships was essential in hitting revenue targets, but only a quarter bothered to track them, and less than 5 percent pursued them strategically.
Stalking Reid Hoffman
The author doesn’t know Hoffman. But Relationship Science quickly showed him many people he knows who could connect him.
If every business is now a relationship business, then of course Goldman hopes that every company will ultimately need Relationship Science. He likes to illustrate that point to prospective customers by taking them on a scavenger hunt through his database. His favorite was the skeptical private equity exec who challenged him to get her kid into a particular private school. Goldman obliged, delivering with one click all the paths between her, her company, and the school’s trustees, as she furiously scribbled them down. Her firm signed on, joining Guggenheim Partners, Jones Lang LaSalle, Perella Weinberg Partners, Nasdaq, Yale, and Duke as just a few of RelSci’s first 300 customers last year.
RelSci suggests the future of networking–and of social networks–will radically reduce the role of luck. If social media today, a decade after Facebook’s founding, presents a warped, cracked reflection of the world around us, RelSci is trying to map that world, especially the professional part of it, with increasing detail, precision, and context. One crucial early step–although possibly an unsettling one–is to collectivize our individual Rolodexes. RelSci already does this: Companies with multiple subscriptions can run searches against all their users’ connections, vastly expanding the likelihood of extracting some value from the network. The more rainmakers whose frontal lobes are uploaded to the cloud, after all, the more paths emerge at that click.
Networking’s next logical step is to move beyond purely public information into richer troves of private data. In one of our conversations, Goldman mused about asking subscribers for permission to passively read their emails in the hope of more accurately gauging the strength of their relationships. And once the algorithm knows you better than you know yourself, the step after that is to let the software quietly mine those “combinatorial possibilities,” engineering serendipity for you–even while you sleep.
As a growing body of research demonstrates, we usually can’t even tell who the most important person in the room is. Not even RelSci has cracked that one yet, but it has come the closest and is still improving. One feature, for example, is a helicopter view of entire companies as the sum of their relationships, and how those relationships lead to people in other companies and industries. Where are their connections strong? Where are they weak? And where are the strategic opportunities 95 percent of corporate executives overlook? If they can’t see them, perhaps RelSci’s algorithms can.
Goldman’s focus on fine-grained “medium data” is what sets RelSci apart from its biggest and most obvious competitor, LinkedIn. With more than 277 million profiles worldwide, LinkedIn is nearly a hundred times RelSci’s size and intends to get bigger. Its long-term goal, according to senior director of data science Jim Baer, is to build the “economic graph,” a totalizing representation of “every opportunity in the world, and every worker, every school, and every entity.”
LinkedIn’s size is RelSci’s opportunity, as expansion drives LinkedIn’s increasingly iffy signal-to-noise ratio even lower. Casual users like me roll their eyes at the random friend requests and misleading endorsements that are hallmarks of the site these days. “None of that shit,” mutters Langone. “I have guys emailing me every third day with LinkedIn requests.” (Users will fluff anything.) Still, not everyone is on LinkedIn. Goldman says that half the names in RelSci’s database have no presence on social media.
Former Thomson Reuters CEO Tom Glocer, also a RelSci investor, believes LinkedIn and other large public gardens will get better at filtering, and that more niche competitors will inevitably spring up. (Bloomberg, for example, has long talked up the idea of a “people index.”) But Glocer is confident RelSci has a three- to five-year lead, thanks to what he sees as the depth and elegance of its database.
Goldman expects that most customers will eventually experience the service as an integrated feature of Salesforce (another investor) or some other CRM system. But that’s thinking small. Imagine adding Google Glass to the picture, with a RelSci-powered facial recognition app matching the wearer’s profile against people in the room, suggesting targets for sucking up to, along with the optimal pathways.
If RelSci’s model of reality has a crippling flaw, it’s in encouraging users to assume that the name on the door is the one with all the juice. Mapping true influence isn’t that simple. Ronald Burt is a sociologist in the University of Chicago’s Booth School of Business who for more than 30 years has studied the phenomenon of “structural holes,” i.e., gaps within organizations. Those who bridge these holes, he has found, produce more ideas, make better decisions, and prosper accordingly. But they aren’t necessarily the ones in charge.
To demonstrate, Burt brandishes a network map of one of the largest pharmaceutical companies in the world. Among its top executives, he explains, a single person is all that ties the group together. The irony is that he’s a relative nobody–several tiers below the CEO and the heirs apparent. Burt’s point is that the boldfaced names are not always the ones we should be chasing. “RelSci doesn’t answer the question of ‘why them?’ “He says. “It assumes you know.” That assumption, he makes clear, is often wrong.
The combination of ubiquitous sensors, Big Data, and social networks means we’ll soon be able to spot and seal structural holes in something close to real time, in line with Goldman’s dream of fracking our networks’ full potential.
MIT’s Alex “Sandy” Pentland is the leading prophet of relationship science, which he calls “social physics.” By understanding how ideas and information flow, Pentland says, “you can reinvent organizations to make dramatically better decisions.” A few years ago, his team in the Media Lab’s Human Dynamics Laboratory invented “sociometric badges” that record a person’s movements, posture, and conversation patterns. Using these badges, organizations can not only tell who’s working together, where, and for how long, but also how seemingly banal measures like face-to-face conversations correlate to performance and creativity. He also found the natural “charismatic connectors” who straddle Burt’s structural holes.
If Burt and Pentland are correct that the social dynamics of business are becoming subtler and more formalized, the inevitable next step is to try to make them automatic. A pioneer in this area is a Silicon Valley-based startup named Ayasdi, which has developed an entire subfield of mathematics–topological data analysis–that renders any Big Data set as a network derived from hidden patterns. CEO and co-founder Gurjeet Singh calls the emerging surprises “digital serendipity.” The software works just as well examining promising drug compounds or cancer research, but the most interesting uses are in social networks. Just ask Ayasdi’s incubators, the National Science Foundation and DARPA.
Maybe the clearest and most dystopian glimpse of where all of this is headed comes courtesy of Tim Hwang, who breeds artificially intelligent “socialbots” as co-founder and chief scientist of the Orwellian-sounding Pacific Social Architecting Corporation. In 2011, he and his co-founders trained fake Twitter accounts to tweet at several thousand humans, deceiving them into tweeting back at the bots and eventually at one another. When the experiment ended, they found the bots had effectively wired their human targets together.
Now Hwang is teaching his socialbots to translate online influence into real-world influence. He’s working with clients in public health and politics to put his experiments to work, whether that means fighting disinformation about vaccinations or selling a candidate. In the case of the latter, that entails identifying the most influential supporters in a target’s circle, then nudging them via socialbots to steer their friends on a particular political race or issue.
All that’s holding socialbots back “is that [online] social networks haven’t become important enough to established politicians and policymakers,” says Hwang. “But as an entire generation grows up on these networks, you’re going to see a much broader impact.” After social media comes social engineering.
Goldman is way ahead of Hwang: RelSci’s data scientists are already filling in the next quadrant on its map of the Establishment: Washington lobbyists. Following his speech in Philadelphia, I bump into his principal supplier. Bruce Brownson runs a database called KnowWho, specializing in politicians all the way down to the county level, with histories going all the way back to high school. His data power Facebook’s and Microsoft’s own lobbying campaigns. “We’re all lobbyists now,” he says, “we just won’t admit it. We lobby people while watching our kids on the ball field; we lobby in the PTA.”
Brownson has a point. And as Big Data grows inexorably bigger, we’ll spend less time stumbling in the dark, searching for a connection, and more time actually connecting. But there is an obvious pitfall: An algorithmic arms race is already under way. As Glocer, the ex-Thomson Reuters CEO, puts it, “the danger is that the tools get to be so prevalent, everyone is on the tools.” And that will be the last time anyone talks about forging a genuine human bond.
Even now, using RelSci is not without its perils. It’s tempting to do what I did with Goldman in Penn Station–just take whatever names the database offers and throw them around, hoping to capitalize on any legitimacy they provide. But the software doesn’t know everything. The flip side of using public information is an inherent degree of uncertainty. It can’t tell me if my friend Khanna, who travels in the same Davos circles as Goldman, recently crossed him in a business deal–or dated his ex-wife. (At one point, RelSci’s editors considered adding an Adversaries button, before realizing they could never keep up.)
Feeling a little guilty, I finally confess to Goldman that the first words out of my mouth were a lie. He’s more amused than disappointed; pointing out that I only went astray by passing along Khanna’s non-existent regards. A more truthful way of phrasing it would have been simply, “I think you know my friend Parag.”
And that’s more or less exactly what happens with Lux Capital’s Wolfe a few weeks later: Moments after hanging up, he calls back to tell me excitedly we have Khanna in common. We bask for a moment in a sunbeam of engineered serendipity, but I can’t help thinking: I really wish I’d known that before we talked. From now on, I will.