Why Strong Ties Matter More In a Fast-Changing Environment

Who should you go to for the best information? And how much do you want to share your information?

Those are some of the essential questions about knowledge management that continue to be asked both by researchers and by executives. Those questions have become particularly interesting in the past five years with the rise of social media that makes it faster than ever to share information and easier than ever to reconnect with hundreds, if not thousands, of people from your past.

Marshall Van Alstyne, associate professor at Boston University and a visiting professor at MIT, is in the thick of it.

“My original training was computer science, followed in grad school by managerial economics at Yale then MIT,” says Van Alstyne. “I studied information productivity, social networks, and information flows in organizations to see if we could figure out what it is that makes people more productive.”

Van Alstyne ran a five-year National Science Foundation study, tracking information flows and e-mail and what makes individual knowledge workers more productive. He also does research on platform economics, what he calls “the creation of ecosystems involving lots of users and developers such as the platform developed by Apple.” Lately, he’s at work on a new study funded by an NSF grant on the creation of knowledge marketplaces.

“You can think of knowledge markets as crowdsourcing, but our particular twist on it is the economic optimization of crowdsourcing, applying economic theory to these social science properties,” he says. “How can we get better answers, get higher rates of contribution? How can we do better resource allocation? Can we value the information shared? Can we cause economic growth inside an information economy? Those are our areas.”

In a conversation with David Kiron, executive editor of Innovation Hubs at MIT Sloan Management Review, Van Alstyne explains how some of his new research challenges the existing theory about the value of strong ties versus weak ties, and why we should beware of “interrupt-driven communication.”

In the social business survey that MIT Sloan Management Review conducted recently with Deloitte, we saw a big industry variance in the perceived value of social tools. Energy and utilities, for instance, score low, while the entertainment, media and communication industries tend to have a much higher valuation.

I haven’t studied those industry sectors, per se, but I’ll offer some speculation on this. I’m going to guess that you have a relatively stable, non-changing environment in the older industries such as energy, and you have much higher update, much faster clock speed, much higher volatility in the entertainment and information technology industries. When information is changing so fast, it’s the recency of information that may matter more, so you need constant updates. This might explain why social media are more valuable in one industry and less valuable in another.

Let me give you a reason why I think that might be the case. Are you familiar with the weak ties literature and the structural holds literature in social networks?

I am, but it would be great if you could go over it for our readers.

Let me give you some of the original theory, and then describe the work that we have just published in a different context.

The original social network theory went after the question, “Where do you find novel information?” And, specifically, “Can novel information help you find a job? How should you go about doing that?” These are great questions.

The original answer provided by Mark Granovetter [now at Stanford University] was, well, possibly by a weak tie, meaning an acquaintance rather than a friend. The reason was that strong ties are likely to be redundant. They’ll have the same information, so you’re not going to find out about new job opportunities. That is an extremely heavily cited paper. I think it’s got something like 14,000 to 16,000 citations; it’s one of the most famous papers in sociology. [See “The Strength of Weak Ties” (PDF), American Journal of Sociology, Volume 78, Issue 6 (May 1973), 1360-1380.]

Yet Ron Burt [now at The University of Chicago Booth School of Business] subsequently refined that theory with a theory of structural holes, arguing that, it’s probably not the strength of the tie that’s critical. What’s really making the difference is bridging the boundary between two different groups that have different information. It’s bridging a structural hole that might make the biggest difference, and there’s where you might get novel information.

A colleague, Sinan Aral, and I just published research in the American Journal of Sociology that reinterprets a lot of that theory and gets a completely different finding.

Your new research reinterprets Granovetter and Burt? How?

All we do is we re-ask the question in a slightly different way. Rather than asking, “where do you get the most novel information?” we asked, “where do you get the most novel information per unit time?” If the environment is stable, you get exactly the original answer of Burt and Granovetter.

But if, on the other hand, the environment updates constantly, you can get the opposite answer of Burt and Granovetter. In fact, you might want a close, cohesive, strong tie that’s constantly updating you with novel information. We found three reasons for this.

Reason number one is that if the environment changes so rapidly, then you can go back to the same person and learn something new or different. You can imagine stock traders would be a great example. The environment is updating constantly, so their arbitrage opportunities are updating very fast. This element is frequency of change.

Reason number two applies in a renaissance knowledge context. Maybe you’ve got someone who knows so much that you can go back to the same person and talk to them about botany, you can talk to them about chemistry, you can talk to them about politics, you can talk to them about, religion, art and music. In any given time increment, you will not have exhausted the knowledge of a renaissance scholar, and you can go back and talk to them about something different. This second element is breadth of knowledge or the size of the information space.

Reason number three is the uniformity of knowledge distributed over the whole population. If everyone knows the same thing, no amount of structural diversity whatsoever can increase the amount of novelty you’ll receive. In which case, the only thing you care about is bandwidth, which is getting as much volume from a single person as you can during any interval of time. In which case, you’re back to a cohesive, close or local tie.

So, by re-asking the original question slightly differently, “where do you get the most novel information per unit time?” you get a fundamentally different answer than the original paper. We tested this set of theories using e-mail data we had gathered for a study of IT and productivity, and found that yes, you can get that property. The original theory is correct if the environment is stable, the information space is narrow, or the distribution heterogeneous. Yet, the original theory is incorrect if the environment is constantly updating, if the information space is very large and very deep, or if it’s the case that everyone knows the same thing. It’s basically a tradeoff between topological diversity versus high volume bandwidth that determines how much novel information you receive.

That is really interesting.

So just to close the loop, this tradeoff could explain the differing results in your study with Deloitte. It’s likely that the original theory applies in the case of energy and utilities, where the value of social media was low, while the new theory applies in the case of entertainment, media, and communications, where the value of social media was high.

You mentioned stock broker arbitrage as an example when you were discussing frequency of change. I noticed in another paper you have where you mentioned a situation where there’s a fast changing job market. Could you tell that example?

Thanks for reminding me. So, the original question asked by Granovetter was, “where might you find a job?” And Granovetter’s original answer was, maybe from a weak tie, because the information would be non-redundant.

If you look at the broader theory, that may or may not hold. It may also be the case that you have to ask, “Is this information recent, or is it stale?”

Consider two different job opportunities. One is a great opportunity, one is a lousy opportunity. The great opportunity will disappear because it’s a job everyone wants. The lousy job will stay open longer because it’s a job no one wants.

Now if you go to a close tie, a friend who knows you really well, someone who is not necessarily bridging a structural hole, they may be interested in pushing that information to you because they have your interests at heart. Whereas if it’s a remote tie across a structural hole or it’s a weak tie, you may have to go ask them for that information, in which case, by the time you’ve gotten it, it’s already stale.

So it really depends on the nature of the information you need. If it’s information that’s really time sensitive, you may need to rely on people who are looking out for you rather than people who are remote from you. Even in the original context, the idea that your close ties may be better sources of novelty is very important, because in this case it’s time sensitive.

What do you say to the basic resistance some people have to using social media for their ties? We’ve heard this — that knowledge is a source of power, and that opening up and sharing that knowledge can undermine that self-conception.

My answer is to think of the difference between information scarcity and information abundance.

Before the introduction of lots of these technologies, certain information was scarce. You, as the control point for access to that information, would have a lot of power. You would have a lot of benefit.

In contrast, when information is abundant and it’s easy to go around you, you can benefit by being the facilitator. You can benefit by being the broker. You can benefit by being the first to provide that information or by being an easy to work with source.

It’s that information abundance that I think will really shift the mindset. If you’re not participating, someone else may be able to provide that information. Even if it’s not quite as good, someone else can provide it a lot earlier and in such abundance that other folks are still happy to have it. It’s almost like an intellectual property regime where you get the reputation benefit of having been first. If you come in third, fourth or fifth, maybe you have somewhat better information, but, by that time folks have already passed you by.

Where have you seen a shift in this mindset at a corporate level?

One of my favorite stories is from an information sharing environment at SAP. SAP, of course, is the large enterprise resource planning, ERP systems, company.

They built a developer ecosystem where developers can answer each other’s questions. Before the introduction of this system, a value added reseller on top of SAP’s software had no particular reason to help out another value added reseller. As a matter of fact, one might not want to answer the question of a competitor because it might make them more competitive.

But after the introduction of this question and answer marketplace, things shifted completely. Now you earn points in proportion to the value of your answers. Now it’s the case that value added resellers are telling their employees to go in and answer the questions of other value added resellers to prove, hey, we’re the ones with the expertise, not those guys. It’s completely shifted the incentives. Folks are now pushing their information into the marketplace. And it happens in a way that benefits SAP.

It’s a really clever mechanism that completely inverts the incentives from one of hoarding to one of information sharing.

Where are the management lessons from your research and the SAP example that are most useful for a manager thinking about implementing some kind of social software initiative, like something around idea generation?

One is that you absolutely will be able to get better idea diversity by approaching large populations, by crowdsourcing. My favorite example comes from a situation that played out at Princeton University with its student government, but it’s directly analogous for almost any manager.

The student government was trying to figure out how to spend their budget, so they generated about 100 of their own favorite ideas. Then they submitted the ideas to a student vote, but allowed the students to add their own ideas. And three of the top five ideas ended up not even anticipated in the original list of the elected student officials.

So one of the first lessons is that you can use crowdsourcing to not only get diversity, but also to choose priority, and do a fairly good job of that.

Another lesson, of course, is to realize when to apply the old versus new theories. There is a diversity-bandwidth tradeoff. The faster paced the environment, the more strong social networks matter.

You’ve studied e-mail for years, and its role in social networking. Most people can’t stand it.

Well, I like the technologies that give me control, and I’m cautious allowing interrupt-driven communication. E-mail is useful among the different criteria, such as having a conversation back and forth. There are lots of criteria. Can you interact, or is it just broadcast? Can you find what you need easily? Is the search good? Can you connect your communication streams to your task streams? Is your communication flow, be it e-mail or Twitter or blogs, connected to your other information management tools, your documents, your task lists, your co-workers?

All of those — the flow rates, the interruptions, the interactivity — really do matter, and the greater the integration, the easier it is to use any given tool. I mean, in a garden you pick the best tool for the task, right? You don’t plant small, delicate flowers with a backhoe. You do excavate with one when you’re building. Pick the tool for the task. E-mail is good for some things; Twitter and Facebook are good for other things.

I think as everyone has experienced, you can get too much information, you get too many distractions. There’s a marvelous study I came across from some researchers in England with Hewlett-Packard, where they examined interrupt-driven communications. If you allow some of these social technologies like Twitter or Facebook or others to interrupt you constantly, it can dramatically reduce your productivity.

How did they study it?

Well, they ran laboratory experiments. They gave folks tasks or problems to solve, and they found that allowing yourself to be interrupted all the time, as opposed to focusing on the task and barring interruption, was roughly equivalent to pulling an all-nighter. It’s almost as if you didn’t get any sleep at all that represents the relative loss of effectiveness.

You need to batch your time on task. You really need a focused hour to two hours uninterrupted by instant messaging and phones and texts and Twitter feeds.

One point we haven’t touched on that’s worth mentioning is that social tools are also being used a lot by companies to eavesdrop on conversations they never had access to before. Companies are obviously scanning Twitter feeds and Facebook and things like that to see what others are saying about them, so they’re able to eavesdrop on conversations that they couldn’t hear previously, and that’s occurring inside and outside the organization.

That begs another question, which is that the freedom and the opportunity to give voice to your opinions does come with that kind of a downside. The price of freedom is eternal vigilance. Do you have a perspective on that tradeoff?

I do. I think in general, companies need to respect their users more with respect to their privacy and what firms are observing versus what they’re offering.

On the one hand, if it’s a Twitter feed and you know that you’re putting it out there publicly, analysis of any of that is, of course, fair game, and companies could, and probably should, be doing that kind of research.

On the other hand, if it’s private communications, and e-mail is among the channels for which you have some credible belief that you’re sharing it with a smaller, more select group — for example, many folks have elected to share only some of their newsfeeds with some contacts — then for the firm to abrogate that trust and then, via analysis, share that with third parties, is probably an abuse of the privacy. I think firms need to be more vigilant about respecting the privacy of their users on those occasions when they have a reasonable expectation of privacy. I’m already seeing some backlash.

A director at HP research labs told me a great story that her daughter had just re-friended her on Facebook. Since they were already connected, the director asked why her daughter was connecting all over again. The daughter shrugged and said she was tired of being spied upon by all the apps on Facebook so she’d created a new account just to get rid of them.

And yet, often it is the case that customers say they want privacy, but too often they’ll part with their profile information for a discount off of milk and eggs — you know, a twenty cent coupon. So what people say and what they do can sometimes be two different things.