Dec 11

We communicate with four, but consume from many more.

When I talk about how how social networks are structured, certain things always resonate with people. One of these is that although the average Facebook user has 130 friends, they only communicate directly with four of those people in any given week. Direct communication includes likes and comments on their posts, posts on their wall, chat conversations, video calls, and private messages.

People I talk to are always surprised at how low the number is – only four people per week, and only six people per month. What’s more, the majority of people in these small groups remains consistent from week to week – for example, our partner, our closest friends and family. Changes in people from week to week is usually posting a like or comment with a much weaker tie, for example seeing someone we went to school with get married, run a marathon, or have a baby.

When telling this story, I usually gloss over an important related fact. Although the average Facebook user is only communicating directly with four of their 130 friends in any given week, they are consuming content from a much larger number of those people. After all, over 50% of active Facebook users come back every day. If you include consuming updates from people as communication, then people are interacting with many more than four, but much of the communication is asymmetrical in nature. I may not communicate directly with you, but I do keep up with what’s going on in your life.

This isn’t a new phenomenon. In years gone by, we kept up to date through word of mouth – chatting with people in our town or village as we went about daily life. We spoke to a small number of people regularly, and because approximately 70% of conversations are about other people, we learned a lot about many others through interacting with a few. We gossiped, because gossip helps us understand how others behave, and helps define social norms. Many of the same motivations exist in consumption of others’ updates online.

Jul 11

This is just the beginning

Disclaimer reminder: I currently work at Facebook and worked on Google+ up until the end of 2010. This post does not reflect anything I did at Google, or anything I’m doing at Facebook, and is simply my personal opinion about the state of the world.

Since Google+ launched last week, many people have been asking me my opinion about it. Unfortunately I can’t talk about specifics (hello, non-disclosure agreements) but I can talk broadly about the state of the world.

When it comes to representing relationships online, there are two big questions:
1. Our offline relationships are very complex. Should we try and replicate the attributes and structure of those relationships online, or will online communication need to be different?
2. If we do try and replicate the attributes of our relationships, will people take the time and effort to build and curate relationships online, or will they fall back to offline interactions to deal with the nuances?

We’re only at the beginning of trying to answer these questions. Google+ is a well designed product, but it is not “the solution” to the problem of representing complex relationships online. In fact, there probably isn’t “one solution”.

If you think about the first question above, Google+ is both trying to replicate offline social network structures (with circles) and build social network structures that are unique to the online world (with following, and with the fact that anyone can add anyone to a circle, independent of whether these people have met offline). Is this the best approach? No-one knows. If history has taught us anything, it’s that trying to predict the future is a fools game. Especially when that future is wrapped up in complex relationships and network effects. Remember, this is just the beginning.

The second question is the big unanswered one. Most user experience problems can be defined with the simple equation: Is the effort I need to go through worth the perceived benefit? Is the effort of creating circles, and managing them over time, worth the perceived benefit of sharing to those circles? Is the effort of figuring out who is in the audience of someone else’s circle worth the perceived benefit of the value derived from commenting? Again, no-one knows the answer to this question. But it’s going to be fascinating to see it play out.

Finally, it’s worth noting a trend that will make the task of representing relationships online even harder. Many fields of science are starting to discover that most of our behavior is driven by our non-conscious brain, not by our conscious brain. This refutes much of our understanding of how the world works. When we meet people, for the first time, or for the ten thousandth time, there are far too many signals for the conscious brain to take in, analyze, and compute what to do. So our non-conscious brain does the analysis for us, and delivers a feeling, which determines how we react and how we behave. It’s our non-conscious brain that will be deciding which social network succeeds and which one fails. It’s going to take most, if not all, of our lifetime to figure out what is happening in the non-conscious brain. This is just the beginning.

Nov 10

Kik and creating a sense of place

You can walk into any bar on any street and immediately make conclusions about whether it is for you. How bright or dark is it? How clean or dirty is it? What’s on tap, Budweiser or India Pale Ale? What’s on the walls? Who is in there? What age are they? What are they wearing? The list goes on. These are some of the things that make up an environment, and they give us signals about whether this place is for us.

One of the most important things when designing online social experiences is to consider how the product decisions you make contribute to the “feel” of the place you have built.

Take the example of Kik, which has seen impressive growth since launch by anyone’s standard. When signing-up, Kik aggresively went through your contacts, picked out people already on Kik you may have communicated with at some point in the past, and put those people right in your Inbox as suggestions. Putting aside the glaring lack of informed consent (that’s a post for another day), it seemed like Kik was trying to help get you started by connecting you with people you may know.

However, my experience with Kik was that I could only recognise about half those people. And as the suggestions have continued coming in, I’m starting to recognize none of them. Having people I don’t recognize as a suggested connection gives Kik a certain atmosphere, shaping the environment. It’s less like dinner at a friend’s place, and more like the anonymity in a sweaty heaving nightclub. This is a place where it’s encouraged for strangers to connect.

This aggressive suggestion system may have fueled the impressive Kik growth, but it has also determined the Kik environment. It’s the wallpaper, the choice of furniture, and what’s on the menu. And looking through the comments on the Android App download page gives you a sense of that environment:

If Kik is a place for strangers to give their vital statistics, how could it possibly be a place for connecting with your closest friends? Watch this space to see how Kik evolves, and think carefully about the decisions you are making, and how they are shaping your products environment.

Oct 10

What do early adopters of social web applications look like?

I’m sure we’re all familiar with the model that includes early adopters and laggards. Some people embrace your new product early and evangelize it to their connections. Others wait, and adopt it once there is a critical mass.

Many think of early adopters as having certain characteristics. We think of them as people who have a tendency to embrace new ideas early, people who look for new product announcements, people who are recognized by others as trendsetters. They tend to be young and male. They like technology.

I’m beginning to wonder if this model breaks down when we think about the social web. One of the first activities one undertakes on many social systems is to add connections. But early adopters on new social systems don’t add others because they are also early adopters, they add people because of an affordance set by the service. For example, on a music sharing service people add others who they want to share music with. On a family connection service, they add their parents and relations. On an event planning service they add people they want to meet face to face. Their first degree connections are people they want to communicate with.

The likelihood of early adopters’ first degree connections having the classic early adopter characteristics listed above are low. We probably need a complete rethink of Everett Rogers’ model for the diffusion of innovations.

The two people who adopted on day 2 have very different characteristics. You can imagine how this effect would cascade.

Jul 10

The data behind The Real Life Social Network

Many people have asked me about some of the references for my Real Life Social Network talk. So here they are. I’m truly standing on the shoulders of others. For the most part, I’ve taken other people’s research and synthesized it, looking for patterns and trying to figure out how it all relates together. I hope the links here inspire you as much as they have inspired me.

Mapping people’s real life social networks.
I published a research paper in 2007 that detailed an early version of this process. I’ve since iterated on it a few times. The paper also contains some findings towards the end.

The magic number 150.
See this New York Post article where Robin Dunbar describes how different groups are made up of 150 people. Nicholas Christakis and James Fowler have also studied this in modern groups. For a great overview (with data) of Dunbar’s number and online games, see this blog post by Christopher Allen.

Strong and Weak ties
Wikipedia provides a good overview of the research literature on strong and weak ties. The seminal paper is Mark Granovetter’s “The Strength of Weak Ties.”

We have a small number of strong ties
In their book Connected, Nicholas Christakis and James Fowler describe one study they conducted with 3,000 Americans. See also research conducted at the Center for the Digital Future at the University of Southern California.

Average number of friends on Facebook
Various research shows that the average number of Facebook friends ranges from 120 to 180. For two examples, see “Rhythms of Social Interaction: Messaging Within an Online Social Network” by researchers at HP Labs, and “Social Network Activity and Social Well-Being” by researchers at Carnegie-Mellon and Facebook. Various research shows that almost all friends on Facebook are people that users first met offline. For an overview, see “The Problem of Conflicting Social Spheres” by researchers at Manchester Business School. For interacting with small numbers of our friends on Facebook, see “User Interactions in Social Networks and Their Implications” by researchers at UC Santa Barbara.

Phone usage and strong ties
Most of this data is from ethnographer Stefana Broadbent. See her presentation at the TED conference. Broadbent has done much research into how people communicate with each other. You can follow her work at usagewatch.org. In particular, see the article “The small size of our communication network”.

Usage of communication tools
The Pew Research Center have much research into this topic. For examples, see “Teens, Cell Phones and Texting”, “Social Isolation and New Technology”, “Social Media and Mobile Internet Use Among Teens and Young Adults”, and “Twitter and Status Updating”.

Different types of friendships
For a detailed look at empirical research on friendships, see the book Rethinking Friendships by Liz Spencer and Ray Pahl.

For an introduction to cognitive biases, see this Wikipedia article. For further detail check out this full list of social cognitive biases. The fact that we make decisions based on our limited information is part of a theory called bounded rationality. The Tipping Point is nicely summarized on Wikipedia, including key ideas and challenges to those ideas. In their book Connected, Nicholas Christakis and James Fowler describe how mutual best friends are most influential, how three degrees of influence works, and the concept of hyperdyadic spread. Other research papers that I reference frequently are “Identifying Influential Spreaders in Complex Networks” by multiple researchers at Universities in the USA, Israel and Sweden, and “Effects of Word-of-Mouth Versus Traditional Marketing: Findings from an Internet Social Networking Site” by Michael Trusov, Randolph Bucklin and Koen Pauwels.

How hubs work
In his book Six Degrees, Duncan Watts explores high and low thresholds for idea adoption, how hipsters influence within a network, the analogy of seeds in nature, and his studies on lists of music. Two research papers that influenced me on hubs and adoption are “The Role of Hubs in the Adoption Process” by Jacob Goldenberg, Sangman Han, Donald Lehmann, and Jae Weon Hong, and “Opinion Leadership and Social Contagion in New Product Diffusion” by professors at Wharton and the University of Southern California.

Multiple facets of identity
danah boyd has done some amazing research over the years, a lot of which relates to identity. For example, see “Profiles as Conversation: Networked Identity Performance on Friendster” by boyd and Heer. Ben Gross has also conducted some great research, see “Addressing Constraints: Multiple Usernames, Task Spillage and Notions of Identity” by Ben Gross and Elizabeth Churchill, and “Names of Our Lives”. Another good paper to check out is “Trust and Nuanced Profile Similarity in Online Social Networks” by Jennifer Golbeck.

Anonymous ratings
See the research paper “I rate you. You rate me. Should we do so publicly?” by researchers at the University of Michigan.

Awareness of Privacy
The following three research papers are a great place to start: “Information Revelation and Internet Privacy Concerns on Social Network Sites: A Case Study of Facebook” by Young and Quan-Haase, “Reputation Management and Social Media” by the Pew Research Center, and “How Different are Young Adults from Older Adults When it Comes to Information Privacy Attitudes and Policies?” by researchers at UC Berkeley and the University of Pennsylvania.

People underestimating their audience
See the research paper “Characterizing Privacy in Online Social Networks” by Krishnamurthy and Wills.

People misunderstanding privacy settings
Multiple research studies show how people misunderstand the privacy implications of their activities. For examples, see “Strategies and Struggles with Privacy in an Online Social Networking Community” by Strater and Lipford, “Expandable grids for visualizing and authoring computer security policies” by researchers at Carnegie Mellon University, University of North Carolina, and Gonzaga University, and “How Different are Young Adults from Older Adults When it Comes to Information Privacy Attitudes and Policies?” by researchers at UC Berkeley and the University of Pennsylvania,

So, that’s a lot of links and a lot of research, happy digging!