Science of Social blog

Can Social Technologies Increase Our Dunbar Limit?

By MikeW

Can Social Technologies Increase Our Dunbar Limit?

by Lithium Guru on ‎11-05-2010 02:55 PM - last edited on ‎03-03-2014 11:57 PM by Community Manager Community Manager

Dr Michael WuMichael Wu, Ph.D. is 927iC9C1FD6224627807Lithium's Principal Scientist of Analytics, digging into the complex dynamics of social interaction and group behavior in online communities and social networks.


Michael was voted a 2010 Influential Leader by CRM Magazine for his work on predictive social analytics and its application to Social CRM.He's a regular blogger on the Lithosphere's Building Community blog and previously wrote in the Analytic Science blog. You can follow him on Twitter at mich8elwu.



Welcome back. Sorry for my digression on Social Network Analysis (SNA) last week. I really thought that was a good introduction to SNA and really helped highlight some of the historical development in the field. For those who have watched it, I hope you learn something. Now back to Dunbar’s magic number!


Previously, we examined Dunbar’s number in great detail. By understanding how Prof. Dunbar estimated his famous number, we can identify two potential mechanisms for potentially raising the Dunbar limit: the number of stable relationships that we can maintain.

  1. First, you can maintain more relationships by reducing the attention devoted to each. This, as we’ve already seen from an earlier post, will tend to result in weaker ties strength on average.
  2. Secondly, we can also make socializing more efficient, so it reduces the time and attention needed to develop and maintain our relationships.

We’ve talked about the first mechanism in an earlier post, and this time we will discuss the second. Since this post is a continuation of the previous, I recommend skimming through the following posts if you haven’t already read them.

  1. The Relativity and Economics of Relationship
  2. Where is the New Dunbar Limit?


Socializing vs. Communicating

Last time I briefly described another hypothesis of Dunbar. It claims that the evolution of a larger human neocortex that enabled language processing has made socializing more efficient so that we don’t have to spend as much time and attention on social grooming. This is what pushed the Dunbar limit for humans to 148, which is substantially higher than other primates that have smaller average group sizes.


It is therefore very tempting to make the following argument: If technologies can make socializing more efficient, so it reduces the time and attention needed to develop and maintain our relationships, then we would have more attention resources to spend on extra ties. Thus, we can only expect modern group sizes to increase. But can modern social technologies really make socializing more efficient? Let’s examine this more carefully.


First, I must emphasize again that “social” is not a revolution! Humans have been socializing for thousands of years. In fact, even chimps and monkeys socialize via social grooming. Modern social technologies arise from a communication revolution. If you look back, I’ve been careful to use the term socializing, as opposed to communicating when I talk about building and maintaining relationships. We have changed the way we communicate, but ultimately not the way we socialize. Let me explain what that means.


Although the internet has made communication more efficient in many ways, it is also limiting in other ways. The internet cannot transmit many nonverbal signals (e.g. touch, scents, physical proximics, body languages, etc) that are often very important for relationship building. Psychologist knew all along that nonverbal signals are extremely important part of social communication. Research suggests that nonverbal cues are 4.3 to 13 times more effective than verbal cues. Moreover, people who have difficulty sending or receiving nonverbal signals often have serious problems with interpersonal and social relationships. Some of these problems are so debilitating that psychiatrists have considered them disorders (e.g. autism spectrum disorders and Asperger syndrome).


Language Evolution vs. Communication Revolution

So what is the difference between language evolution and communication revolution? The difference is that the evolution of language is an addition to our previous means of socializing. It is an additional channel of information to the other nonverbal cues. However, modern communication through the internet is more like a replacement of our previous mode of socializing. When you communicate through the internet (by making heavy use of the newly evolved verbal channel of language), you’ve completely eliminated all the nonverbal channels.


Even if you use voice and video, the subtle details that your sensory system can pick up from a face-to-face engagement is far greater than the Internet’s current capacity. The fragrance, the sense of touch, the eye contacts and feedbacks, they all tell you something about a person that you cannot put in words. Finally, just ask yourself a simple question. If you want to establish a relationship with someone, would you prefer to only socialize with him/her through the internet, or would you eventually want to meet the person in real life?


Since social technologies are not able to help us build relationships more efficiently, I must come to consensus with Prof. Dunbar that modern social technologies probably cannot increase our Dunbar limit any further. Last time we discussed the tradeoff between the number of relationships we maintain and the strength of these relationships. To some extent, this is the limit we have to work with, because our attention resource (emotional intensity and time) is limited and hasn’t change in the recent evolutionary history of mankind. Therefore, in order to maintain more ties, we either have to lower the overall tie strength (see figure 1b), or simply maintain fewer number of super strong ties (see figure 1c). In either case, the average tie strength is weaker. However, the blue-green area under the curve (which represents our total attention resources) remains the same, and it is equal to the yellow area in figure 1a.




Social Technologies Provide Greater Accessibility

Despite the somewhat disappointing finding (that social technologies cannot help us maintain more relationships), technologies do bring something new to the table. Two points deserve special mention here. First, it helps us overcome the limitation of distance and time. With social technologies, we can now meet like-minded people from all over the world who speak the same language. And we can talk to them anytime. Thanks to the asynchronous nature of communications on Twitter, Facebook, communities, and social networks, we have easy access to a whole new world of people that we could never reach before.


Second, social technologies can help us maintain accessibility to a huge number of ties that are so weak that they would ordinarily be forgotten or lost in real life. Social networks help us maintain these ultra-weak ties because social data are persistent and searchable, so we can always search and look up the forgotten engagement with these acquaintances. These ties are not the stable relationships that Prof. Dunbar talked about. Personally, I would not even consider these real relationships, because I almost pay no attention to them. It is like having a few extra pages of names tagged onto the bottom of my address book that I never bother to look at. Since these ultra-weak ties take virtually no attention, we can have many of them in our online social network. But that doesn’t mean they are real relationships until we spend the time and attention to develop them.


The accessibility brought about by social technologies is very valuable, because it gives us more choices on how we allocate our limited attention resources. In both cases, it lets us choose from a much larger pool of candidates to whom we want to spend time with or pay attention to. So we can spend our limited attention resource on people that are most interesting to us and matter most to us.



We have covered a lot today. So let me briefly summarize everything:

  1. Socializing is more than merely communication. It includes many non-verbal cues that are extremely important for building relationships.
  2. Social technologies are not likely to increase our Dunbar limit further because they cannot help us socialize more efficiently. So we can only maintain more relationships by redistributing our limited attentions resources.
  3. The value that social technologies bring is the ease of access to a much larger population, so we can make the best use of our limited attention resources on people we care about.

Alright, now that we have a deeper understanding of Dunbar’s number from the attention economics perspective, we are ready to use it to explain some long observed phenomena in social media. Before I reveal the topic for next week’s discussion, let me ask you a few questions:

  1. Does your company or brand have a FB fan page?
  2. Does your work involve working with a FB fan page?
  3. Do you find it hard to engage your FB fans?

If you answered YES to any of the above questions, then next week’s post should be very relevant to you. So stay tuned and watch out for my twitter feeds. In the mean time, I welcome any questions, comments, criticisms and suggestions as usual. Or we can just discuss on the science of social.


And don’t forget next week’s social customer virtual summit. If you are planning to attend, please be sure to come by either my live chat sessions (which is in the agenda) or head over to the Networking Lounge if you want to chat with me. I’m always happy to talk. If you have not registered, you can still register now. It is FREE. I hope to see you there.




by Daniel Alfon(anon) on ‎11-08-2010 08:29 AM



There is a recent question I saw on Quora and I was wondering if you could comment on it and share your thoughts about it.Perhaps you'd like to shade light on the confusion some have about the essence of social networks, personal social networks, as well as strong vs. weak ties.Here is the question:



An example is if you have informed your social circle you need a job and it yields no results, you have the greatest chance of success asking a complete stranger on the streets. Such an individual has no overlap with your own social circle, and could thus provide an entirely new network to tap.




by Lithium Guru ‎11-08-2010 10:48 AM - edited ‎11-08-2010 06:32 PM

Hello Daniel,


Thank you for directing this excellent question here. For those who are not on Quora, I will repeat the question here in addition to the example you quoted. The question is


"In theory, the strongest social network is one where no one is connected. Have any social networks taken this into consideration, if it not, how could they?"


The problem is that the inquiere never gave a clear definition of what a strong social network is. Simply using the adjective "strong" to describe social network can lead to a lot of confusion and misleading result. Before I answer it let me give a little bit of historical context first. I believe this arise from the argument that "weak ties" are much more efficient for information propagation because it enables info from one clique (i.e. a densely connected group of strong ties to  each other) to reach distant cliques of strong ties. And this is a very well established result from Prof Granovetter with solid HR data backing it up with your job search example you cited.


The context of the posed question is the entire social network, not someone's personal social network, which is the correct context we should be talking about ties strength discussion. Because there is a relativity of tie strength a work here (see The Relativity and Economics of Relationship). Strong relative to what? Weak relative to what? One person's weak tie may very well be another person's strong tie. So tie strength is usually relative to someone's personal social network.


When you look at someone's personal social network (e.g. network of people who are connected to 1 person, say A), there are always a continuous distribution of tie strength raning from the weakest ties (acquantiance or even strangers) to the strongest ties (families and close friends). But we like to draw a line somewhere and call tie strength beyond that line strong ties, and those below that line weak ties. But where people draw that line seem to be pretty arbitrary.


Now, within A's personal social network, the strong ties tends to connect among the close friends of A, so information do not propagate beyond A's clique where it was generated. But since weak ties tends NOT to connect among the close friends within A's clique, these weak ties can propagate information beyond the clique (beyond A's personal social network) much more efficiently. Because these weak ties are strong ties to some other personal network (that is how you are able to tap into the weak ties network). Note that strong ties and weak ties a relative to A's personal social network here.


Now, the problem is that the person who posted the question made the conceptual leap that weak social network means no one is connected to each other. That is unfortunately not correct! When no one is connected, you have no network, and no information propagation at all. Using the example that was quoted, if you ask a stranger for job in a social network in a world where no one is connected, then you have near zero chance of getting a job, because that stranger is not connected to anyone else too, so he have no network for you to tap into. In such a world, even the definition of stranger is ill-defined, because no one is connected, so basically everyone is stranger.


In Granovetter's original paper, weak ties are defined to be those with little overlap among their social circles. (This may not be 100% corrected with the tie strength in interpersonal relationships which is what Dunbar's number is about.) But there has to be a tie, and there has to be stronger ties in the personal social network (if there are no stronger ties within that personal network, that that weak ties would be the strongest tie, because there is nothing stronger). So the weakest ties are network bridges, (a.k.a. connectors, boundary spaners, etc). That is, the weakest ties in someone's social netowrk are ties that overlaps at one 1 person, not 0 person (i.e. disconnected). Because even though they connect to your personal social network with only 1 person, they have their own social netowrks who are more densely connected to themselves. It doesn't mean they only connect to 1 person in the entire social network. See the problem. People simply mixed up personal social network and The Social Network, which usually implies the entire social netowrk consist of overlapping personal networks (see my series on Cyber Anthropology). Everything is relative, that is where the confusion arise.


So, to answer your question. The strongest social network (here it means the entire social network) is NOT one where no one is connected. The strongest social network is one where everyone is connected to everyone else, so that everyone is already part of everyone else's personal social network. However, due to Dunbar's limit, our personal network cannot include everyone. So we must have strong ties and weak ties within our personal network. And because a weak tie in my network is usually a strong ties in some other network, that is why they are more efficient for info propagation outside of my own network. It is not because they are weak ties in the entire network and not connected to anyone.


OK, I hope I have clarify some confusion here. If not please feel free to ask me further questions. I always respond to questions on my blogs.


Thanks again for commenting and for asking such a great questions. Hope to see you around more on Lithosphere.


 In theory, the strongest social network is one where no one is connected. Have any social networks taken this into consideration, if it not, how could they?

by ryanglasgow(anon) on ‎11-08-2010 06:37 PM


Thanks for your thoughtful response. I was the one who asked the original question, and have reworded it as follows:

If your personal network cannot fulfill a need, you have the best chance of asking someone you have very little or no relationship with. Have any social networking websites taken advantage of this principle, if it not, how could they?


An example is if you have informed your social circle you need a job and it yields no results, you have the greatest chance of success asking a complete stranger on the streets. Such an individual has no overlap with your own personal network, and could thus provide an entirely new network to tap.



by Lithium Guru ‎11-08-2010 11:05 PM - edited ‎11-08-2010 11:09 PM

Hello Ryan,


Thanks for rephrasing your question and posting it here.


What you describe is what Granovetter call "the strength of weak ties." And we are taking advantage of it all the time. Only this kind of social structural (one that you are referring to as "social network") is really not a social network. It is really just a community. Please see my post on the difference between Social Networks and Communities. A "social network" where few people know each other is by definition NOT a social network.


What holds this social structural together is a common interest just like a community. You are seeing the benefit of it right now. We do not know each other and are not on each other's personal social network. But because of our common interest in networks and communities, we can help each other. And in this case, because you posted this question to the mass public (which is a very large population). So your chances for getting an answer is rather high, because it only take 1 person who know the answer to help you. In this case, that person is me.


I strongly recommend you take a look at my series of blog posts on Cyber Anthropology. Especially the 3rd article in that series: From Weak Ties to Strong Ties, which discuss this precise phenomena that you mention about crowdsourcing for job opportunities.


OK, I hope I've address your question. Please let me know if you like to explore further.


Thanks again for the discussion and hope to see you around.


by Frequent Commentator on ‎11-11-2010 10:33 PM

Hello Michael,

Another great read and as always interesting viewpoints that provoke new pathways to thinking.


Excellent point on the distinction between socializing and communicating. I have personally sometimes treated 'networking' as even a third category overlapping these two. The thought being that when someone is networking, they are adding nodes to their network <b><i>in anticipation</b></i> of a future exchange/benefit but the tie might for all practical purposes remain inert for a long time. It is also possible that among some of those you network with has a larger overlap with your immediate interests and needs and so you end up socializing/communicating with them depending on the context. Anyway, just some musings.

Also agree with you about weak ties being more efficient for information propagation. As you point out, they might be  a weak tie from your point-of-view but a strong tie with other networks. Also, it is also possible that some of these weak-ties are part of (albeit weak) many different domains/networks and so are privy to a broad spectrum of surface events that bubble up within each network. And this also makes them a good vehicle for info propagation between networks (and great domain-bridgers).

Now to a question -- this is in regards to your statement<i> "Since social technologies are not able to help us build relationships more efficiently..". </i>

We both agree that attention & time impact tie-strength. I also agree that technology cannot replace face-to-face engagement. However, couldn't social technologies help us in relationship 'evaluation' and help us decide where we want to develop the relationship further using face-to-face and which ones we are happy to have just a technological exchange with? Doesn't this make socializing more efficient?





by Lithium Guru ‎11-12-2010 11:58 AM - edited ‎11-12-2010 12:20 PM

Hello Ned,


Welcome back. Thanks for the comment and the kudo.


There are a lot of things that people do, pretty much every social verb that we can think of in a group context can be a category of its own: socializing, communicating, networking, entertaining, collaborating, co-consuming, value co-creating, etc. All can be a category of its own. I simply pointed out the difference between communicating vs socializing because people tend to mix these up.


Certainly networking can be a important category of things that we do, especially these days where our society is infused with social technologies that enables us to establish weak ties and ultra-weak ties so easily. I've heard people call these ties relations, rather than relationships.


Weak ties in someone's personal social network are almost surely going to be strong ties in some other people's personal network (those people may not be people you know, but they exist). Because everyone has some strong ties. Although their strong ties may not be as strong as your, but tie strength are relative. So if you look at the distribution of ties strength for anyone, they will span a range. So there will always be a strongest ties in any person's personal social network. And these strong ties will affect their behavior and therefore affect the information propagation.


Finally, to your question. Social Technologies can definitely help us evaluate and decides to whom we can to develop the relationship. This is the accessibility that I mention in my blog. And it definitely enables use to evaluate, and choose which relationship we want to develop further. So if you want to consider selection of ties as part of socializing and relationship building, then YES, social technology can help you do that more efficiently. But for building interpersonal relationship in terms of building tie strength between 2 persons, then IMHO, the answer would be NO.


Anyways, thank you for the comment and question. I'm sure a lot of people have similar inquiries as well, so I'm very glad that you brought it up. See you again next time.


by Frequent Commentator on ‎11-12-2010 06:19 PM
Hi Michael, I am with you on the number of different categorization possible and my mention of networking was more to highlight a point that you can be 'linked' with someone without either socializing or communicating. And it really comes down to as you said weak-ties.

I would however like to challenge you a bit more on the social technologies and building relationship pov. You are unequivocally stating that social technologies does not help build interpersonal relationships in terms of tie strength. And I think it does contribute to building relationships - but definitely not to the extent as in a face-to-face engagement.

As you mention in your post, internet cannot transmit many nonverbal signals like touch, scents etc. You are absolutely right. But compared to the pre-social world of print, chat, email, phone to the current world of Facebook, YouTube, Twitter etc., I do think that the addition of these channels have allowed relationships to form 'faster' and reach the threshold faster before engaging in a face-to-face. The scenario that comes to my mind are the dating sites where folks first meet online - but then the relationship is taken to the next level through the use of social technologies - before eventually meeting face-to-face.

Anyway, would like to hear your thoughts on this. Wanted to push this exchange one more level :-).

Also, I would generally agree with you that eventually folks who want to establish a strong relationship with someone do desire a face-to-face. However, as we have talked before it will all come down to how one is defining strength and strong-ties. I have penpals who I have corresponded with over 20 years without meeting them face-to-face. Are they weak ties - definitely not. Are they strong ties - it depends :-).

Always enjoy the chats Michael.

by Lithium Guru on ‎11-12-2010 07:16 PM

Hello Ned,


Thanks for coming back to carry the conversation further.


I believe my orginal thesis was more correct than my quick wording in my reply to your previous comment. What I want to say is that social technologies do not help us build relationship (in terms of interpersonal tie strength) more efficiently. The key is more efficiently, but more efficiently than what?


Since the context was about Dunbar's number, which is really only valid for pre-industrial ages (pre-social media era), because Dunbar's conclusion were extrapolated from data from that time. So I am comparing how people build relationship now (during the social media revolution) to pre-industrial age, where people typically can only rely on face-to-face engagement to build relationships. That is the comparison I want to make.


So my point is that social technologies cannot help us build relationships more efficiently than face-to-face engagement. And I believe we are definitely in agreement there.


The main thesis of ths post is that although social technology help us gain access to much large population (which help us evaluate, decide and choose where, and to whom we want to spend our limited attention), it cannot replace face-to-face engagement. In fact, any of the digital medium, email, chat, etc. (that you mention in your resonse) are no comparison to face-to-face interaction. Modern social technologies are definitely doing a lot better than those older digital communication media, but that was not the comparison that I was trying to make. I apologize for not making this clear in my earlier response.


Until we have technologies that is far more advance than the holodeck, the most efficient and effective method for relationship building is still face-to-face engagement and real-life interaction.


Well, if you eventually meet the person in real life (IRL), than all the verbal communication on social platforms becomes an addition rather than a replacement to the face-to-face engagement. That is why it seem to enable relationship to form faster. Sure, if you combine face-to-face with social technology, then of course that is better than face-to-face alone. But again that is not my point. I am comparing the scenario between face-to-face vs social technologies. So if you already have a relationship with someone IRL, then social technologies are definitely great for maintaining that relationship, because you get the benefit of both face-to-face and online social interactions.


With your example of a 20 year penpal: Take a look at how much attention (emotional intensity + time) you have devoted to this person. That is the extent of your tie strength between you two. But remember, tie strength are somewhat relative. If you just meet a friend you have not met for 20 years, you action and behavior around this person may look like you have very strong ties with him. A real measure of ties strength is really the amount of attention you spend on the person. Whether you call it strong tie or weak tie depends on where you draw the line along the continuous spectrum of tie strength.


OK, I hope this clarifies the confusion.

Thanks again for the discussion.


by Frequent Commentator on ‎11-12-2010 08:49 PM

Hi Michael,
Thanks for the clarification and quick response. You hit the core point with "A real measure of ties strength is really the amount of attention you spend on the person" -- totally agree. What I was trying to process is how much of that 'amount of attention' breaks up between online vs face-to-face and how does that break-up contribute to the tie-strength.

Also, for a second I want to switch from efficiency to effectiveness. Irrespective of whether social technologies help in building relationships more efficiently, I definitely think they contribute to building more effective relationships. Given our schedules and the limited time we have in our hands (not to mention the location constraints), each of us can only have a limited face-to-face engagements with each other (outside of family, immediate circle, work etc.). Social technologies help us level set our perceptions, expectations, and mental models in a way that when face-to-face engagement does happen, there is a higher likelihood of success in sustaining that relationship and making it stronger; as opposed to a face-to-face meeting with very little prior interaction.

Thanks for an interesting convo.


by Lithium Guru on ‎11-13-2010 12:21 AM

Hello Ned,


Thanks again for the discussion.


On the topic of efficiency vs effectiveness of social technologies: I would still say that social technologies are less efficient than face-to-face engagement. As you said, they do contribute, so they are not completely ineffective. So they are effective, but not efficient.


For example, to build a relationship up to a certain level, it may only take a few face-to-face meeting, but may take months of online social engagements. From a different perspective, you can think of a control experiment where you have only 8 hour to spend to get to know a person. In one scenario (A) you can only meet the person face-to-face, where as another scenario (B) you can only communicate through social technologies. Then after exactly 8 hour, you can ask the question: which case, A or B, would build a stronger relationship? I believe that in most cases, scenario A would build stronger relationship. Notice that I say most cases, there are always special edge cases where scenario B might build a stronger relationship, but they are rare.


In case B, you can definitely learn something about the person and build some kind of relationship. So it is NOT ineffective; it is does have some contribution and therefore effective. But it is just less efficient, because you spend the same amount of attention resource and get a weaker relationship out of it. That is the main reason I would use "less efficient," when comparing engagement through social technologies vs in-real-life.


Alright, I hope I made this point clear. Let me know if you still disagree. I'm always happy to discuss further. This would only make both of us learn more. So never hesitate to add further comments. Kudos to you too for being so perseverant.


by Frequent Commentator on ‎11-13-2010 09:33 AM
Hi Michael,
First of all let me clarify something - I am not really disagreeing with you :-). On the contrary, I agree with your general conclusions. What I was trying to probe was the extent of effectiveness or ineffectivess when using (a) social technologies alone, (b) face-to-face alone, and (c) social technologies along with face-to-face and the impact on relationship building as one goes from one end of the continuum to the other.

As to the control experiment, I agree thaat in most cases (A) would build stronger relationship in your scenario. Now if you were to add a third scenario (C) where they still have 8 hours but the first 4 hours you can only use social technologies and the next 4 hours you engage in face-to-face, my hypothesis is that (C) will in the long run outperform (A) -- this is the effectiveness I was talking about in my last comment. Of course, I might be proven wrong and I still have to gather empirical evidence on this.

Here are some of the reasons why I think (C) will outperform (A) in the long-run.

  • Face-to-face engagements are impacted by personality factors. Sometimes folks get tongue-tied when meeting someone for the first time and it takes time for the relationship to ramp up. Social technologies gives a sense of comfort for some people to speak up while not being 'exposed'.

  • Face-to-face communication requires instant response. While this is good in some ways, it also means that many a times the response given is not the most ideal one. Soc.technologies on the other hand can provide the person with some 'thinking room' and they can provide a more adequate and appropriate response to the other party (less chance of a negative response or response that might be misunderstood)

  • You talked about non-verbal cues that help in relationship building. Sometimes these very cues are also good at destroying relationships or nipping off a relationship before it had time to develop. Our faces, moods, and expressions sometime reflect a prior event and at times the other person wrongly attributes those cues to the current conversation.

  • Anyway, circling back to my first statement - you have made valid points in your post and comments. The most important point being that social technologies today cannot replace face-to-face engagements. They may augment it in some form or fashion. The second point being that social technolgies help us make better use of the attention economy.

    Thanks Michael - I learned a few things from this post for sure :-)

by Lithium Guru ‎11-13-2010 10:55 AM - edited ‎11-13-2010 11:01 AM

Hello Ned,


Welcome back again. I understand. And I don't think there is really any big diagreement between us either. But it is just interesting and fun to carry the discussion deeper to a more abstract level. So please don't feel in anyways that there are disagreement that must be resolved. Moreover, everyone is entitled to their own perspective. And you definitely have a lot of valid points.


Despite the absence of empirical data, I think it is definitely possible that your scenario (C) (i.e. 4 hours of social technologies + 4 hours of face-to-face (F2F)) could out perform scenario (A) in the long run. The key is "in the long run." But what happen right after the 8 hours? Do you still think that scenario (C) would result in a stronger relationship than (A) right at the end of the 8 hour? In the long run, pretty much anything can happen. 


I definitely agree that sometime F2F engagement could destroy a relationship faster too. But that is actually a result of the fact that F2F is a more efficient form of social engagement. We can learn more about the other party in a shorter amount of time, whether it is the fact that we like them or dislike them.


Also, I generally agree with you that certain sense of privacy and not being exposed can definitely help people speak more frankly and honestly. However, I like to point out that really comes down to a motive problem. It is no longer about relationship anymore. If someone is saying something that he doesn't want people to know that he said it, then I don't think the speaker has the intention to establish and develop the relationship at the first place. The intent and desired to further the relationship is the first requirement for any relationship (see How Do People Become Connected). In this case, we are comparing apples to oranges again.


I think whether social tech can help us build stronger relationship really comes down to how we use it. That is whether we are using it as an addition or replacement to F2F engagement.


For example, I use social technologies to learn about my friends and families in Taiwan. But due to the fact that we are geographically segregated, we can probably only meet F2F once a year. With or without social tech, we will and we can only meet once a year for about a week. In this case, if I use social tech, it will be an addition to our year F2F gathering. So the additional amount of time/attention I spend on these distant relationships through social tech would probably increase our tie strength. But if we are using it as a replacement to F2F, the result would not be the same.


Finally, I must reassure you that I really do enjoy these type of conversation. Guess what? I am really an academician at heart. And I really think that making our thought process visible and searchable through these debate/discussion here could help a lot of people learn about this subject too. It is a win-win-win situation (i.e. for you, for me, and for everyone else). So I'm always glad to carry on the conversation further, and I am always learning from these discussion as well. So thank you.


OK, have a great weekend and see you next time.


by Larry Irons(anon) on ‎11-16-2010 09:06 AM



I enjoyed this discussion, especially your highlight of the "thickness" of copresent communication when compared to using social technology. I would also suggest, drawing from the conversation analysis tradition of research in Sociology, that the difference also involves the speed of the interaction. As my dissertation advisor, Deirdre Boden and Harvey Molotch said in their essay "The compulsion of proximity":


"Copresence is efficient not just because so much information is present at any given moment (e.g., body talk, position of an utterance, speech particles) but also because it allows individuals to move rapidly through streams of turns, as each turn quickly leads, iteratively and implicatively, to the next" (p. 267).





by Lithium Guru ‎11-16-2010 05:56 PM - edited ‎11-16-2010 05:57 PM

Hello Larry,


Thank you for stopping by and commenting.


Definitely co-present communication (where all our neural sensory pathways can cooperate to give you sort of a whole Gestalt of the individual that you are interacting with) is a very powerful thing. I am, however, not very familiar with the conversation analysis that you mentioned. Regardless, I certainly see your point about the speed of interaction can lead to a very different outcome. Very interesting idea.


I will definitely check out the reference link you provided. Unfortunately, page 267 where your quote were extracted from is not available in this Google Book preview. I guess I would have to check out the library next time I go to a University.


Thank you again for the comment and hope to see you again enxt time.


by Matt Kammerait(anon) on ‎11-18-2010 08:24 AM

How do we get beyond the relativity of network positioning?


Because we constantly have to define tie strength in terms of another factor (a tie is strong for the objective of job seeking, or product recommendation and purchase behavior) - can there be a cohesive picture of an individual players role in a given network? 


I feel like a hierarchy or factor based definition might be a step in the right direction - something like Meyers-Briggs. Some sort of analysis along these lines would be interesting - i.e. I'm a low clustering, high betweeness, low closeness, high degree member of a given network which would correspond to a given role and specific strengths and weaknesses of my position.

by Lithium Guru ‎11-19-2010 12:59 AM - edited ‎11-19-2010 01:02 AM

Hello Matt


Glad to see you here. Thank you for stopping by and posting this interesting question.


The standard way to get around the relativity of network position is through a combination of various network metrics on the nodes, not the edges. So this is qutie different from the tie strength concept that was talked about here, even though there are also network metrics on the edges as well.


Tie strength is an intrinsic property of the tie between two entities (nodes). They are NOT typically defined in term of other factors. We should not talk about ties strength for the objective of job seeking. In fact, Granovetter, who discovered that weak ties are better for job seeking, still calls these weak ties. They are just better for job seaking. The strength of the tie between 2 persons are independent of what objective we are trying to achieve. If you try to define tie strength in terms of these other factors, than the notion of tie strength will become arbitrary. For example, someone who picked up your airline ticket, which you carelessly misplaced at the cafe can all of a sudden become a very strong tie if your objective is getting on the plane. You can call these strong ties for getting on the plane, but I wouldn't do that.


However, there is something else that network science usually refer to as the importance of a node, and that is defined relative some objective. Some nodes (people on your social graph) are better for job seeking, other nodes are better for WOM, yet others are great for something else. I think it is the notion of importance that you are thinking about when you talk about network positions. Because there are many ways that a node can be important to a the network, that is why there are many different kinds of network centrality measures (Degree-, Betweenness-, Eigenvector-, Closeness-, etc). Aside from centrality measures, there are many network structural metrics, such as Clustering coefficient, Core number (see all the ones we compute in this post). And some of these network metrics (including centrality, structural metrics, etc) do depend on the tie strength, such as Eigenvector centrality.


So back to you question. The standard way to deal with the relativity of network position (or relative importance of a node) is through the use of different network metrics. Those who have high degree centrality, may not have high betweenness centrality. These nodes are important for networking, but not good for job seeking and innovation. Those who have high degree and high betweenness, may havffe low closeness. These nodes are good for networking and innovation/job seeking, but not good for speedy propagation of information. So relative to different objective, you will get a high or low score depending on what the objective is.


That is why there is no universal network metrics. You have to look at a collection of these network metrics to get a holistic understanding of why a certain node is important and in what ways, relative to what objective. If you only look at one metric, like Degree centraltiy (aka, the number of connections), you can only know one way that this person can be important. You know nothing about how important he is for innovation. For that, you have to look at the Betweenness centrality.


And people like Marc Smith have define roles by a collection of network metrics. But they are just a name, a place holder for these network metrics. For example Answer Person = high out-degree, low in-degree, low clustering. It is simply a place holder for these set of network metrics with this particular characteristic, which are important in some ways, but are not important relative to other objectives. This really has nothing to do with tie strength, because it is an attribute of the edges in a graph, where as importance is an attribute of the nodes, although some (but not all) node metrics depends on the edge attributes.


Alright. I hope I've clarify more than confuse people here. Since you ask a rather tehcnical question, I thought it is only fair to give a little more technical and deeper response. Let me know if anything is still unclear. I'm always happy to discuss further.


Thanks again for the question, and hope to see you more on Lithosphere.


by Matt Kammerait(anon) on ‎11-19-2010 05:55 AM

Thanks for your thoughtful response Michael! 


From a practical standpoint, and an actionable one, do you find its better to focus on the importance of a node when looking at differential incentives? The balance I'm having trouble understanding is how granular to get in my analysis of a given network (i.e. customer network, prospect network, trade show activity network, etc.) in determining how to stage response. If I go too far down the rabbit hole in looking at different potential node categories then I'd end up needing as many versions of the incentive as there are nodes - on the other hand, if I stick with two (or a few) distinct categories like "influencers" and "general nodes" then I don't feel I'm doing justice to the network data that's available. Is there a happy balance to be found somewhere between those two extremes?


Pardon my initiating this discussion on this post - I know it's something of a tangent. Perhaps there's a more pertinent comment section where we can converse?


Side Question: Is there a list of Marc's network roles ("placeholders") somewhere?

by Lithium Guru ‎11-19-2010 11:48 PM - edited ‎11-20-2010 12:16 AM

Hello Matt,


Welcome back and thank you for carrying the conversation further.


What you’ve described is a very common problem with any complex data sets, and it is nature to social network analysis. This is because human relationships are very complex. Every relationship is unique in some ways, but in our modern social networks platforms, we collectively call these relationship friends, connections, etc. And we treat them as the same, where they should be different. That is why it feels not right to just have a few small number of categories. It doesn’t match our intuition, because we have so many different kinds of relationship and each one is valuable in its own way. But as you said, if you really want to find the unique value in each node, you will have so many that it is not actionable.


I don’t know of a happy balance between the two extremes. But my way of dealing with this is not to pre-specify a set of categories a priori. Rather, I will figure out what problem are we trying to solve first, then use as many categories as needed to solve the problem. If the problem can be solve with 2 categories of nodes, then great. I will find the network metrics that distinguishes these two categories. If the problem is harder and requires 5 categories of nodes, then I will find the collection of network metrics that are necessary to distinguish all 5 categories. As a scientist, I must say that we should not prescribed to any fixed solution. We should always look at every problem as a new and treat it with objectivity. And then we should try to look for the simplest solution that have the greatest explanatory power.


As for Marc’s network roles. You can find the ones he defined in this paper: Welser, H., Gleave, E., Fisher, D., & Smith, M. (2007). Visualizing the Signatures of Social Roles in Online Discussion Groups. The Journal of Social Structure, 8(2).


We have subsequently identified more roles in our communities using the 10 network metrics we compute along with other participation metrics. I won't go into what metrics define them, but the roles include:

1. networkers,

2. influencers,

3. trusted person,

4. trend setters,

5. connectors (aka, network bridge, boundary spanners, etc),

6. inquirers (aka questioners),

7. experts (aka answer person),

8. enthusiast (aka fans),

9. creators (aka originators),

10. commentators,

11. critics (raters),

12. moderators, and

13 collectors.


If you only look at a few metrics, many of thse would not be distinguishable. But if you have hundreds of metrics, then all of a sudden they can be easily distinguished. Because even though many of the metrics are the same, there will be a few metrics that separate them apart.


Alright. I hope this addresses your question. Thanks again for the discussion and hope to see you next time.


by ToddStark(anon) on ‎12-30-2010 07:20 PM

Some of Dunbar's recent comments on the Dunbar limit with respect to social networking.



by Lithium Guru on ‎01-01-2011 10:20 AM

Hello Todd,


Thank you for the link.


I totally agree with Prof Dunbar. Being a neuroscientist researcher in my previous life, I can totally understand how our brain limits our relationship building capacity.


Thanks again for commenting. See you later.