Designing for Social Interaction

Posted by

It took both the telephone and the mobile phone 15 years to amass 100 million users, but Facebook did it in 9 months. We see more and more people becoming connected on online social networks, and it seems our networks are growing exponentially. But the reality is, social networks rarely add to our number of connections. We’ve already met almost all the people we’re connected to on social networks. We’re already connected to these people offline. Social networks simply make the connections visible. For example, we often connect with old school friends, and catch up over a couple of wall posts. But rarely do we continue the conversation once we’ve connected, and over time we forget that the connections exist. In fact, Facebook users often have no interactions with up to 50% of their connections.1 When we study how people are interacting on social networks, we see that most interactions are with a very small subset of the people we’re connected to.

The average number of friends on Facebook is 130, and many users have many more.2 Yet despite having hundreds of friends, most people on Facebook only interact regularly with 4 to 7 people,3 and for 90% of Facebook users, 20% of their friends account for 70% of all interactions.4 We also see this with phone usage. We have hundreds of people in our phone contacts, yet 80% of phone calls are made to the same 4 people. We know dozens of people who use Skype, yet 80% of Skype calls are made to 2 people.5 Even when people play computer games online, they mostly play with people they know offline.6

We also have varied interactions with the people we’re not as close to. We find out about new jobs, not through our friends, but through friends of friends. When our friends’ friends’ friends lose weight, we lose weight.7 We go onto eBay and buy things for hundreds and thousands of dollars from people we’ve never met, and will probably never interact with again.

We have many diverse relationships with the people in our lives, yet the web doesn’t support this very well.

the web doesn't support different layers of relationships very well

On Facebook (left), all my “friends” are treated equally. I’m presented with a long alphabetical list. Some of these people I would trust with my deepest secrets, and there are others that I’ve met less than five times. Yet they are grouped together in a big bucket of “friends.” Of the people who are “friends” with me on Facebook, there are many that I wouldn’t call a “friend.” I may call them an acquaintance, or a colleague, or even a family member, but they’re not in the same category as my closest friends.

In my phone (right), all my contacts are treated equally. I’m presented with a long alphabetical list. My best friend is given the same number of pixels as someone I haven’t spoken to in 5 years. I may be able to access my most frequently contacted people via a ‘Favorites’ tab, but these people are still presented in the same way as the people I’ve lost touch with. On LinkedIn (right), all my connections are treated equally. Yet I have worked with some of them every day for years, and there are others that I met once at a conference.

Our social web tools must start to understand the strength of ties, that we have stronger relationships with some people than with others. And with this knowledge they need to adapt.

There are three kinds of relationship ties:

  • Strong ties: People we care deeply about.
  • Weak ties: People we are loosely connected to, like friends of friends.
  • Temporary ties: People we don’t know, and interact with temporarily.

Let’s look at each type of tie, and how we might design for them.

Most people have less than 10 strong ties

For decades, people have talked about social networks being made up of strong and weak ties. Think of the people in your life. Think about your closest friends, the people you are closest to in your family. These are examples of your strong ties. Strong ties are the people you care about most. People often refer to strong ties as people in their “circle of trust”. Strong ties often wield the most influence over people‚Äôs decisions.

Most people have very few strong ties, usually less than ten. A study of 3000 randomly chosen Americans showed that the average American has just four close social contacts (four strong ties). Most Americans have between two and six strong ties. People’s strong ties come from a variety of places. About half of the strong ties are said to be friends. The other half includes spouses, partners, parents, siblings, children, co-workers, fellow members of clubs, neighbors, and professional advisors and consultants.8 In both 2002 and 2007, a study of 1,178 adults found that on average, people had about 10 friends they meet or speak with at least weekly (10 strong ties).9 So, when we’re designing for strong ties, we’re designing for small groups of people.

Some people believe that this is changing, that the web is making us closer to more people.10 On the contrary, research studies have shown that the vast majority of usage on social networks is between strong ties. As we saw earlier, on Facebook it’s with 4 to 6 people, with phone calls its with 4 people, and with Skype it’s 2 people. When people play online computer games with others, they are mostly interacting and playing with people they know, often with people who live less than a few miles away. This pattern of technology being used for strong tie communication is not new. When the telephone was invented, it did more to expand and strengthen strong ties than to weaken them. A study in the 1970s showed that the majority of phone calls were to people who live within five miles of the caller’s home.11

When designing for strong ties:

  • Think about their existing means of communication. Phone calls, text messages, email. Strong ties already have established ways to interact, we should support them, and not try and replace them with our own messaging systems.
  • Showing more information about the ten closest people is likely to be much more valuable than showing less information about many more people.
  • Avoid generic terms such as “Friends”. This will likely lead to over-populating groups and reducing their relevance.
  • Suggest connections to people, but communicate the effects of adding new connections.

People have about 150 weak ties that they stay up-to-date with

Moving away from your strongest ties, think about some of your friends’ friends. People you don’t know so well. These are examples of your weak ties. Weak ties are people you know, but care less about. These are the people that are loosely connected to you. Weak ties are people you wouldn’t feel uncomfortable stuck in a elevator with, yet don’t feel much emotional closeness to.

In the 1970s, the sociologist Marc Granovetter wrote a seminal paper about the strength of weak ties. He concluded that weak ties are often a much better source of information than strong ties. As our strong ties are a very small circle, weak ties can be a more powerful source of information and advice. Studies show that most people can only stay up-to-date with 150 weak ties in real life. This pattern has been true for thousands of years. Neolithic villages tended to separate into two once they reached 150 inhabitants, the Roman army was split into groups of 150, so that everyone knew each other.12 It is still true today, online and offline. We are connected to many more than 150 weak ties, but don’t stay in touch with them. We may be connected to hundreds of people on Facebook, but we would struggle to tell anyone what is going on in all of their lives.

In social networks, weak ties can sometimes be very useful, for example, connections on LinkedIn can help you find a new job, a friend of a friend can give you advice on a specific topic. Yet sometimes, weak ties on social networks can lead to awkward social situations, for example, receiving unwanted Facebook invites from people you don’t know very well.

When designing for weak ties:

  • Consider the trade-off between communication and trust. Weak ties may be more knowledgeable about something we’re interested in, but we may trust them less. It may be important to show our other shared ties, or expose their sources of knowledge, so that we can increase the trust between people.
  • Make it easy for people to expose their networks to people they trust with that data. This will open up links between weak ties, without compromising user privacy.
  • Enable appropriate communication channels between weak ties. It may be better to go through, or highlight, a shared strong tie.

The web is increasing our interactions with temporary ties

Strong and weak ties are not enough when we think of relationships online. We need a new category of tie – the temporary tie.

Temporary ties have always existed, but the web is bringing them to the fore. Think about some people you’ve only interacted with once. You don’t actually know who they are. A store assistant, a call center employee, the person you bought from on eBay. These are examples of your temporary ties. Temporary ties are much more common online than offline. They are people that you have no recognized relationship with, but temporarily interact with for a specific reason. Once the task has been completed, temporary ties are unlikely to interact again. You don’t know these people beyond the words they typed, and whatever online profile they have. With the rise of online user generated content, temporary ties are becoming more important.

Following are four common types of temporary ties:

  1. People sometimes interact with temporary ties around an information need. People needing information seek people with knowledge of the answer. Once the request for information has been fulfilled, interaction with these ties usually ceases. You may ask a person in the street for directions, or online you may seek information from people on Yahoo! Answers. This type of temporary tie is increasingly important for the future of web search. People are looking to other people, rather than businesses, to answer certain queries.
  2. Temporary ties also exist when people need to temporarily interact to complete a task. Once the task is completed, interaction ceases. For example, interacting with a sales assistant in a shop, or having a plumber in your house to fix a leaking pipe. Many of these tasks are now online, for example interacting with a call center representative to arrange cable installation, or buying something from a temporary tie on eBay.
  3. Temporary ties can form around a shared ongoing interest such as a sports team or hobby. Interaction can often happen on a community website, for example a sports team forum. The true identity of these ties can remain unknown. One some of these sites, frequent posters start to recognise each other and each other’s behaviour. Although they likely have never met, their relationship can move from temporary tie to weak tie. There are an increasing number of examples where temporary ties interact online, and meet offline.
  4. People often form temporary ties with people sharing the same physical space. These ties can be as simple as a 1 minute conversation on the street, or as complex as an intermittent 3 day conversation at a music festival. Technology now allows us to communicate with temporary ties who shared the same physical space with us, albeit not at the same time. We can use our phones to see who has previously been to the restaurant we’re considering, what they ordered, and whether they thought it was any good.

Because we have no relationship history with our temporary ties (unlike with our strong and weak ties), representing identity and reputation are critical for successful interactions. We need to know that the eBay seller is trustworthy. We need to know that the medical advice we’re reading is from an actual doctor, and not someone pretending that they are a doctor. We need to know that the restaurant review is from someone who knows about food, and that the Amazon review wasn’t written by a company employee.

reputation building

On Yahoo Answers, people can build their reputation in different ways. On the left, “Messykat” is building her reputation by trying to provide the best answers to questions. She may be helpful in 20% of cases, but we still don’t know much about her credentials as an expert in any of her 3 specialty areas (Cats, Dogs, Weddings). On the right, “RuthAnn” is trying to build her reputation by stating that she “has been training dogs for 40 years”. This only adds a superficial layer of credibility, as we can’t verify if it is true.

building trust by positive feedback from temporary ties

On eBay (left), “jmjenkins” is building trust by getting positive feedback from temporary ties who have dealt with him in the past. Some of these people have also left positive comments on his profile, e.g. “Item just as described, arrived safe and sound. Thanks! A++”.

On Amazon (right), “W.Todd Dominey” is building his reputation by providing helpful reviews in 665 of 711 cases. He is also building trust by using his real name, and sharing where he is from. However, as with “RuthAnn” on Yahoo Answers, we don’t know how much “W.Todd Dominey” knows about the topic being reviewed. We can’t verify his credentials.

The motivation for most people who help their temporary ties is not monetary. You can’t create a great temporary tie community with financial reward. Yelp tried to do this by paying people $1 per review, and later they tried to build communities by paying people $15 an hour to comment on existing reviews and write new ones. Yelp ended up with a lot of poor quality reviews 13 14, and a lot of bad press. Some of the actual motivations for temporary ties include recognition of being an expert, altruism, and feeling that they belong to a community. All of these behaviors need to be understood and accounted for in our social web designs.

Supporting temporary ties is good for business. One study showed evidence that “an increase in positive comments about a mobile handset typically appeared a month or two before an increase in market share for that handset.”15 Another study over 7 months showed that when the online promoter activity (an estimate of the online chatter likely to lead to a recommendation) went up, sales went up the following month. When activity went down, sales went down. 16

When designing for temporary ties:

  • Prioritize a great system for building reputation. Allow people to give feedback to one another.
  • Encourage people to expose content that will increase trust in their identity. This could be their real name, a real photo rather than an avatar, or proof of their qualifications.
  • Prioritize a great system for building trust between people. This may be highlighting shared connections, shared groups, or shared interests.
  • Don’t incentivize people with money, incentivize them to build their reputation.


Social web design will become an important part of every interaction designer’s skillset. To do it well, we’ll need to understand some basics about human relationships. If your users’ needs center around strong ties, you’ll design something very different than if they center around temporary ties. Understanding the difference between strong, weak, and temporary ties will help us build better online social experiences.


1, 4 User Interactions in Social Networks and their Implications, Wilson et al. 2009


3 The Economist article quoting stats from Cameron Marlow, sociologist at Facebook

5 Data from a TED talk by researcher Stefana Broadbent

6 PBS article by Henry Jenkins, MIT Professor

7, 8 Nichola Christakis + James Fowler “Connected: The surprising power of our social networks and how they shape our lives” Little, Brown and Company 2009

9 Reports from the Center for the Digital Future, University of Southern California

10 Brave New World of Digital Intimacy

11 I. de Sola Pool, The Social Impact of the Telephone, MIT Press, 1977

12 Quoted from studies by Robin Dunbar

13 Paying People to Yelp

14 Yelping for Dollars

15, 16 Charlene Li + Josh Bernoff, Groundswell, Harvard Business Press, 2008


  1. There are huge opportunities to increase adoption and engagement in these strong, weak, and temporary ties. Unfortunately, in many of the off-the-shelf social or collaboration platforms I’ve seen (my experience is limited to enterprise social tools), the ability to recognize and target interactions based on how well people may know each other is a pipe dream.

    Um, check that: a pipe dream would suggest some of these providers are aware of the opportunities that exist in those relationships. In fact, I find many of these providers/vendors/etc expect us to be wowed just to be able to add *friends* or *colleagues* because Facebook and LinkedIn do it. Hopefully as these products continue to mature and react to more innovative entrants into the market, we’ll see greater functionality in reputation scoring and relationship management.

    Until then, without sounding like I’m giving up, I feel like my hands are tied to the technology my clients have agreed to long before UX types like myself enter the picture.

  2. Thanks for the comments!


    “…the ability to recognize and target interactions based on how well people may know each other is a pipe dream.”

    I don’t think it is a pipe dream. Many companies are working on building tools that extract social graph data and make that available to others. Many research studies have been conducted on Twitter users and usage by using their API to access their data about who is connected to whom.

    “I find many of these providers/vendors/etc expect us to be wowed just to be able to add *friends* or *colleagues* because Facebook and LinkedIn do it.”

    I think it is up to us to educate others about different kinds of relationships. We need to show people how having one big friends group doesn’t reflect life offline, and presents as many problems as opportunities. It is up to us to push this and get people thinking about doing more than what exists today. You may be tied to a certain technology now, but we need to make sure that people are informed enough so that they might make different decisions next time.

  3. Awesome post with great insights Paul!

    I think the single most critical point you have made here is to separate “weak” and “temporary” ties. The failure to realize this distinction lies at the heart of most critiques of any community driven endeavor. Traditional communities, that the word hark backs to, were people with weak ties. The internet communities are one of temporary ties.

    What internet has really facilitated is the emergence of these temporary ties because the marginal cost of communicating with anyone new is close to zero. In earlier days, the cost of discovery itself was a significant barrier to entry for these communications to happen.

    Again great post, and hope to read some more soon.

  4. Paul,

    I’m very interested in this concept, and I think you’re right about how different design and UI approaches are best suited for particular strengths of ties–particularly professional ones.

    I actually heard a program (Spark podcast) on this recently and began to think about the different types of connections I maintain online. Rather than thinking of them in terms of strength, I thought more about how frequency indicated value. I then mapped it out and found that I had a longer list of connections with whom I spoke every few months than those with whom I spoke several times a month (not including family, friends or co-workers). The people that were in the “every few months” group presented a lot of value to me, though! (They’d probably fit within your category of temporary or specific need-based ties.) I wrote up a blog post on this and included a visualization of my connections map: As I pointed out there, the two to three people that had a significant impact on my career in the last year were in the infrequent group–something I wouldn’t have realized without mapping it out.

    The reputation points aspect of Boxes and Arrows commenting functionality is a perfect example of ways that UI can help to leverage and indicate the value of weak or need-based ties. I’ve actually made a handful of long-term professional connections by participating in discussions attached to posts like yours with people I would probably never have encountered otherwise.


  5. Thanks again for the comments!

    @Sriyansa Many people use weak ties to describe our interactions online with people we don’t know. I’m glad that my distinction of weak and temporary ties is useful for you, it has certainly helped me in my work.

    @Chris I think there is a great opportunity for us to rethink how we represent our relationships in our digital tools. Imagine if we replaced/supplemented our alphabetical lists with representations like you created in your exercise. We could learn a lot of new things.

    The relationships between strength of tie and frequency of communication is an interesting one, and it comes up in research. There are people in many of our lives that we don’t communicate with frequently, yet would turn to for support in an instant if we needed to. These are also the people that we meet, and despite not having seen each other in many months, it feels immediately natural. People describe it as immediately feeling “like the old times” and “like you’d just seen each other a few days ago”. So frequency is a good signal for strength of tie, but it doesn’t always show us the full picture.

  6. Hey Paul,

    Great piece, while I have read about thsi elsewhere, I think you did a good job consolidating some good data references and putting it in a language most people can understand. I sent it to a few stakeholders and other partners from a project I am working on.

    I am the position of, as Chris Avore put it, addressing the pipe dream. About more than year ago my organization brought an out of the box social enterprise solution thinking it would address their needs. Well it kinda did, but not really. I was hired with some other people about 6 months ago to try to understand better our users. We learned a lot about their real life social interactions and are now starting to educating our business partners on social interaction design. For example we were talking about the subject of ties, and how we should enhance the strong ties (which we don’t do at all) and build up the weak ones (which we kinda do a little).

    Our problem now is that we must customize the enterprise tool or build extra custom components to do what we need. So in some ways I agree with Chris Avore, many enterprise companies are buying products and expecting amazing results. But that is not happening. If their smart they can get a good team int there to come in and straighten things out.

    I think that it is up to us to educate the enterprises as well as the enterprise solution providers. I am hoping that the company that provider our software will look at what we’re doing to learn how they can adopt their software. It may be a pipe dream, but I can at least hope.

  7. Paul,

    I definitely agree–frequency is only one piece of the puzzle. In fact, just yesterday I received a phone call from someone I would have put in the very infrequent category and whom I’d actually never spoken to directly before (only via email/Twitter so far). He wanted some advice and felt that a direct conversation was the best approach, even though our relationship had never included that kind of contact before. I was surprised by his call, and then as we got to talking a bit more, even more surprised by my feeling that talking to this person over the phone was so unusual–especially when 10-15 years ago phone would probably have been the primary contact method!


  8. Great post, well explained and provocative. Reminds me that many years ago we did fieldwork with French mobile phone users and talked about technology and social connectivity. We got some very clear explanations about the difference between “copains” (i.e., buddies) and “amis” (friends), where an ami was for life and where one had far fewer amis than copains. Our French client had a lot of trouble understanding, then, our Japanese respondents who were (at that time, using Print Club, those small stickers) aggregating large numbers of shallow relationships they called “friends.” So you have the nomenclature issue, and you have the cultural issue. Even talking about it was tough, if the frame being illustrated didn’t match the existing frame.

  9. Fantastic post! Thought provoking.

    In the comment stream above, Chris Butler touched up on the frequency of communication as being another attribute of a relationship. Steve Portigal mentions how difficult it is even to describe these relationships in simple terminology (Copains et Amis).

    It is definitely helpful to look at individuals in our social networks as nodes and their relationships as bonds that can be described with a set of attribute; but this type of predicate logic may yield to undesirable results – Just like in the real world. Example: the realization that the bonds of a particular relationship are stronger in one direction and weaker in the other.

    How can we address these types of issues when designing online interactions. Do we simply rely on the individuals to “attribute” the type of relationship? Do we let the system provide hints on the state of that relationship? Food for thought, no doubt.

    Thanks for sharing,

  10. @John “So in some ways I agree with Chris Avore, many enterprise companies are buying products and expecting amazing results. But that is not happening.”
    I’m really interested to hear what their goals are. What does “amazing results” mean to them?

    @Chris It’s interesting how expectations of how communication technology should be used changes over time, and between demographics and cultures. One nice example is how many younger people feel that calling someone on the phone is rude, and that you should send a text message first to make sure they are available to chat.
    Here’s another anecdote: I was in China doing research with a merchant store owner. When we were there, a guy walked into the store and introduced himself to the store owner as a new potential business partner. He had traveled 200 miles to make this unannounced visit. When we asked him why he hadn’t called ahead to introduce himself, he said that the first contact between potential business partners must be face to face in order to build trust between both parties immediately. For him, no communication technology would suffice.

    @Steve. Absolutely agree re. nomenclature. This is also a problem when we talk about “social”. The social web. Making things “more social”. Often, people have very different ideas in their head when they use that word.

    @Khalid I think the key here is how much transparency your system has. Sometimes less transparency for one party is better. Examples where this has been done well include Facebook’s “Ignore” a friend request. People need to know that they can ignore someone without the other person finding out. For many actions people need to take to define relationships, they need the action to remain private. Our real life relationships are asymmetrical, and our online systems need to be designed to reflect that.

  11. Hi Paul, Interesting post – thanks. A very trivial question for you: On the Android address book image at the top of your post there is an an orange clock icon next to Amy Lau. What does this signify?

  12. @Adam, the orange clock icon is to indicate that that person is using Google Talk (instant messenger) and is currently idle. If they are active, the icon changes to green.

  13. Paul: do you know what, if any, of these principles when into the design the Google Buzz?

  14. Dear Paul:

    Well done. Great lead and useful statistics in the opening section of this article. I was not surprised that FB and cell phone users do not correspond with all of their “friends,” but I WAS surprised by the low numbers of people most of us actually call or contact.

    You clearly and concisely explain Fowler and Christakis’s insightful work (Connected is a fascinating read) before extrapolating about the implications for UX design.

    Thank You,

    Eric O.

  15. @Znok I worked on Buzz during development and continue to work on it as part of my job today (I also work on YouTube and other social initiatives). One part of Buzz we worked hard on was making it easy to share with groups of strong ties. We continue to try and improve it. Unfortunately I’m not in a position to talk in detail about it currently.

    @Eric Thanks! We do have very small groups of people we communicate with regularly. This pattern is very consistent in our research. There is also some more detail in this recent presentation I gave:

  16. An IxDA colleague referred me to your article – after reading my recent post on Facebook friend circle size ( Your article fills in the blanks for me!

    Dunbar actually talks about a series of numbers, not just 5 and 150. Are more distinct strength levels beyond strong and weak? I notice anecdotal evidence among my friends who are teachers and journalists, who seem to have a moment of truth when reaching around 500 followers.

    Doesn’t this have huge implication for, say, contact book design? For example, when a user moves beyond 150 or 500 contacts, shouldn’t filtering and removal tasks be escalated to mission-critical status?

  17. Paul,

    One thing I’m curious about is the definition of interaction on Facebook (I know I can look at the cited sources, but I assume it’s wall posts, messages, pokes, etc). The reason I’m curious is because unlike Skype, cell phones (for phone calls) and other media, Facebook can unilateral interaction, which is basically stalking. Knowing what someone is doing, thinking and saying at all times is on some level interaction.


  18. well, I think i know something about social media .. lol .. after I read your post become interesting to realize how much I have to learn .. interesting and good point of view you make here .. sometimes was in my mind Twitter is amazing , now Facebook become number one in all my preoccupations..thank you

Comments are closed.