Response: How Does the Web Define Authority?

The real questions are: whom do we trust, why, and under what conditions is trust transferable?

Chris Brogran asks: How does the web define authority?

First is an important matter of language.  There is a large difference between authority and an authority.  Authority is power formally granted by a position or role.  An authority is some who has power or influence; it’s a matter of trust by others for a given context.  E.g. I trust my doctor to diagnose an illness, but have no reason to trust him on gardening.

There are many reasons we (dis-) trust others, including: shared opinions; length, frequency, and consistency of interactions; and how our peers feel about the individual the given context.  These are all correlated, but frequency and peer opinion have the biggest impact on transferability of trust.

How use doth breed a habit in a man! — William Shakespeare, Two Gentlemen of Verona

Frequency can be no surprise, it’s the underpinning of blanket marketing.  Familiarity bred through repetition.  Ever wonder why you stop on a TV show you dislike while flipping through the channels?  Of course, that could be me making excuses for lousy taste.

Peer consideration is the tricky bit.  There is a lot of research going into this, but there are some great seminal works that are written for non-academics covering the spread of innovation and adoption of scientific principles.

When our peers already have some experience or opinion on a topic or an authority, then our own opinions are strongly colored by these existing opinions.  In fact, as odd as it may sound, given a pattern of relationships where the opinion of all but one individual is known, we can predictably estimate both what the opinion is, and how strong the opinion is, of the unknown person.

But how about when we something brand new to us and our peers?  Frequency plays a big role here too.  If lots of people, even people we know nothing about, say XYZ is a good idea, we’re likely to give it the benefit of the doubt, trusting the wisdom of the crowd.

So, strictly speaking, is (dis-) trust transferable?  Depends.  If trust already exists in our network of peers, yes, and predictably so.  If the context is brand new to you and your peers, there is no trust to transfer, but we use frequency as a proxy in our decisions.

So what does this all mean?  Let’s look at an example: you’re trying to decide whether you agree (trust) what I have written.

I am new to writing about this, and I’m not particularly active in social media, so chances are we don’t have peers in common.  You can google for what others have said about me, but you’re not going to find much relevant to trusting me in this context.  Ultimately, because I don’t have a track record (frequency & consistency) for participation (peers) in this context, I am at the mercy of how similar our outlooks are, and any opinions that may develop in comments.

Implied Social Networks: People In the News with Rankings

As a follow up to this previous post with an image of the relationship between people mentioned in the news, I’ve been asked to provide more detail.

First, why bother at all?  Exploring the implied relationships may tell us about the individuals in question, but certainly provide more context to each of the other topics at hand.  This context not only provides additional understanding the of topic, but can also be a valuable research tool in quickly determining which other topics may impact the one at hand.

What are the relationships shown?  Shown are names occurring in the same news articles, which implies a relationship.  This relationship may a formal relationship, e.g. the working relationship of Bush (8) and Condoleezza Rice (10). Or, the individuals may be related to a common topic such as Michael Phelps (4) and Babe Ruth.

Following are the top 20 names, by centrality, and the number of different implied relationships for each.

People in the News (detail)
People in the News (detail)
  1. Barack Obama   1128
  2. John McCain 902
  3. Sarah Palin 405
  4. Michael Phelps 237
  5. Pervez Musharraf  95
  6. Kwame Kilpatrick  103
  7. Hillary Clinton   270
  8. Bush  158
  9. Joe Biden   218
  10. Condoleezza Rice  107
  11. Steve Jobs  160
  12. John Edwards   101
  13. Clark Rockefeller 69
  14. Britney Spears 122
  15. Brett Favre 65
  16. Bernie Mac  60
  17. Miley Cyrus 70
  18. Bill Clinton   148
  19. Anwar Ibrahim  34
  20. Stephenie Meyer   27

What’s the data set?  A random sampling of news sites including NYTimes, Google, Yahoo!, CNN, Drudge, and the like.

Is this an accurate reflection of news?  I am polling a number of the big news sites, so hopefully it’s not far off.

Any surprises? Miley Cyrus!

Deconstructing Delicious: Merlin Mann

Merlin Mann has a large set of public delicious tags, and I thought I’d take a stab at their interrelation.  By my measure of centrality, his top 20 are:

Merlin Mann's Delicious Tags
Merlin Mann's Delicious Tags
  1. 43folders
  2. domains
  3. tumble
  4. music
  5. sanfrancisco
  6. macosx
  7. flickr
  8. mbwideas
  9. gtd
  10. movies
  11. design
  12. tv
  13. selflink
  14. mac
  15. heh
  16. email
  17. productivity
  18. lifehacks
  19. the_man
  20. cigars

Why Merlin Mann you might ask?  Well, I like his work, and he has a walloping collection of tags.

Want your tags drawn and quartiled?  Leave a comment or drop me a line at erich@howweknowus.com…

Predict Attention in Social Networks

People distribute attention according to a power-law distribution.

Power-laws have long been associated with distribution of quantity of links individuals in social networks have. My on-going research suggests that power-laws not only describe distributions at the network level, they also describe distribution at the individual level. We communicate in a power-law distribution with our contacts, by frequency. Initial analysis also suggests we spend time communicating with each other according to a power-law.

The distribution analysis for frequency was conducted across six social networks of various types ranging in size from fewer than 100, to more than 6,000 individuals. Most SN research has been conducted on smaller networks (fewer than 100 individuals); so testing across a wide range of sizes both confirms earlier results and suggests that size is not a factor in the power-law distribution. I was concerned about possible distortion on small networks due to implications from Dunbar’s Number. It turns out that small networks are indeed different, which I am not going to go into here, but they still fit these distributions.

Analysis on any complete sub-set, will still fit these pattern. By complete, I mean that connections between any two individuals in the sub-set, must be the same as in the whole set. The value is the introduction of the ability to sample, and to operate over a network recursively. Similarly, much information can be gained about a larger network, even if the data you have is incomplete.

This distribution may allow us to accurately predict impact of changes to any social network. By measuring the current state, we can estimate the impact of adding/removing people and connections. This could be of tremendous value pursuing in any social goal creating by facilitating cohesion, culture, and the like.

I intend on publishing the results and methodology. If you are looking for that level of detail you’ll have to wait, but mail me (erich at howweknowus.com) if you would like to discuss.

The Never Ending Quest for Data

Luc Legay's Social Network
Radial Representation of a Social Network

Finding good data in this field is difficult, even most of the academic literature references relatively small networks of less than 100 or so individuals. I suggest that the academic research is just starting to take off now (although the field is very far from new), because of availability of large real world datasets available in the social networking sites.

Nathan Eagle (Reality Mining at MIT) was kind enough to share 330,000 hours of proximity and cell phone communications data he and the team collected from volunteers over the course of the project. To say I am quite excited about digging into it, would be an dramatic understatement.

For other large data sets, Duncan Watts is spending his sabbatical over at Yahoo!, and I can only hope there are other people looking really hard at the data available there, Facebook, Hi5, Google, and many more. Research into people’s behavior, especially in a commercial setting is not only a great thing for the unprecedented data, but at least equally as important, this also brings to front the ethical implications.

[Image: Luc Legay‘s Facebook network]

Friendship: #1 factor in whom we spend time with

Mobiles & Communication
Mobiles & Communication

Like all good science, analyzing social networks sometimes works out to proving things we always thought were true. Sometimes, we never even had any idea how right we were. For example, we really do spend more time with people we like.

A few really bright folks from MIT and the Kennedy School, have a paper pending publishing:

[analyzing] 330,000 hours of continuous behavioral data logged by the mobile phones of 94 subjects, and compar[ing] these observations with self reported relational data.

Three significant conclusions:

  1. Self-reported data shows a mildly positive relationship with observed data, but is exceptionally noisy.
  2. Friendship outside of work is the best indicator of who spends time with whom at work.
  3. Physical proximity is a good indicator, and predictor, of friendship (and not-friendship).

So, what do these conclusions suggest for practitioners?

Observed vs Reported Data: Surveys are great for all the reasons surrounding explicit participation, but the bias effects are significant. Find a way to marry active participation with empirical exploration and analysis of social networks.

Friendship and After-hours: Don’t under estimate the power of emotion on business decisions. Since we’re more likely to agree with data that confirms any already held thoughts, let’s be realistic and recognize the impact that, viewed through friendship, has on communication in our firms.

Proximity and Friendship: While I was unable to tease out any correlation/causation relationship from this paper, if we consider friendship as a proxy both for trust and ease of ability to work with (through shared history, goals, culture, etc.), there are some solid implications on the upper limit to the value of outsourcing.

[Photo by Ed Yourdon]

Great Work, Lousy Title

13th Century Social Network of Deeds in France

Good news, from Roland Piquepaille over at ZDNet…

According to Nature News, a team of French researchers has used medieval documents to create the oldest detailed social network ever constructed. The mathematicians and computer scientists looked through thousands of records of land transactions dating back as far as 1260 in a Southwest part of France.

Makes me wonder why I did not come across it earlier. Oh, right, because they titled the paper Batch kernel SOM and related Laplacian methods for social network analysis. Shame on you French Scientists, don’t hide the good stuff.

Playing with Circos

Martin Krzywinski at the Genome Sciences Centre of the BC Cancer Agency, created software called Circos designed to help elucidate the interaction of genes, and has used it to create some truly beautiful graphs.

The software is pretty complex, and I have only figured out how to use his simple on-line version, which limits the number of inputs.  But, even so, here is an image I created using Circos.  The image represents the number of emails exchanged by the top 10 most connected participants with each other, from an active large email list.

Relative % of each other's time
Relative % of each other's time