Musings on Relationship Symmetry in the Enterprise

The title sounds pretty highbrow academic. It’s funny to me.

Anyway, I just read that according to ComScore, Twitter grew 131% in March.


Very true observation Mr. Bailey

That number doesn’t include international or client-based usage. While I read through my feeds today, I got five new follows, all from people I don’t know. Oh, and there’s the whole Aston Kutcher vs. CNN for the most followers race, actually more interesting than it sounds. And the Twitter Easter worm.

Yes, Twitter is en fuego. You can’t hope to stop Twitter. You can only hope to contain it.

I miss Olbermann and Patrick.

One reason why Twitter is so very popular is the asymmetrical follow model it employs. I can follow interesting people, and they don’t have to follow me. Although with celebrities flocking (pun intended) to Twitter, it may become a badge of honor to get a follow from someone famous. Maybe just an @reply?

By the way, didn’t MySpace do asymmetric following first? I’m pretty sure all those bands and movies promoting themselves on MySpace don’t have to befriend everyone who befriends them.

It’s too tough to keep track of which site did what first.

Moving on, Paul tweeted (natch) an interesting post about why Facebook will eventually adopt an asymmetric model. During our initial Connect redesign chats, Rich also pointed me to another post that does an excellent job breaking down the differences between the symmetric and asymmetric models.

I won’t go into detail. Read those if you want more information. The short version is:

  • Asymmetric: I can follow/listen to you without requiring you to follow/listen to me in return, creating four possible relationships. We don’t follow each other. We follow each other. I follow you; you don’t follow me. You follow me; I don’t follow you.
  • Symmetric: To follow/listen to you, I must ask for permission and be approved by you and vice versa, create only two possible relationships. We don’t follow each other. We follow each other.

I prefer to call these friend and follow models because that’s easier to type. Let’s do that.

A while back Rich asked me what I thought about switching to a follow model on Connect. We currently use the friend model, but he and I agreed that the follow model would have value.

For example, a CEO like Jonathan Schwartz who blogs could benefit greatly from the follow model, which works very well for people who broadcast content. The friend model carries too much significance for most high-level executives (e.g. he’s friends with X, but not with me), and there’s a lot more overhead to maintaining a network in the friend model.

The follow model also helps find the mavens within a company because it’s an easy way to gauge how interesting or helpful a person is perceived to be, especially if you subscribe to the low following/followers ratio.

Beyond those two, I think most people in an enterprise would a) not get what follow means and/or b) feel like it was stalking.

Think back to your first go-round with Twitter. Most early users I know signed up and then did very little for a few months because the follow model isn’t as obviously valuable as the friend model. Add to that the fact that many enterprise users may not want to be using your shiny new network anyway, and you may never get their attention again.

Plus, most follow networks do not have real profiles, whereas friend networks do (and must). This raises the creepy factor in a network with profiles, like Connect, when you’re followed by someone you don’t know.

Eventually, we decided to stay with the friend model. This was a few months ago, when Facebook was hot. I think it still is, but you wouldn’t know it. Every other story in my feed reader since January has been about Twitter.

This begs the question would I change my mind now that people are using Twitter?

Not now. The follow model will be very valuable to enterprises in the next few years, as people get acclimated to both friend and follow networks and how they operate differently. I agree that Facebook will move toward a mixed model, and the lessons learned from that will be huge.

It’s going to be extremely tough to blend a network with profiles and the follow model. There’s value there, but will anyone see beyond the creepy factor.

So, for now, we’ll continue to keep both Connect and OraTweet. Since we can control both systems, we may do some lightweight stuff like show the number of OraTweet followers a person has on the Connect profile, or allow a follow from the profile.

We do have a unique vantage over the consumer sites though, allowing for experimentation.

So, what do you think of the two models? Will they blend in the enterprise or elsewhere 🙂




  1. Quick comment – a shift by Facebook from the friend to the follow model would be a major cultural change for Facebook, which has to date emphasized real relationships. Then again, Wal Mart used to trumpet “we buy American” and then shifted its model in a heartbeat, so presumably Facebook could do the same.

    Regarding enterprise applications, it would depend on the company and the way in which it chose to broadcast information. While my former employer had internal social media tools, it chose to broadcast via e-mail. If, however, the CEO of a Fortune 500 company chose to use the internal tools to communicate with his/her employees, a follow model would make more sense. However, if the internal tools are primarily used for collaboration rather than broadcasting, a friend model makes more sense.

    Oh, and not to brag or anything, but Barack Obama followed me on my old Twitter account. 🙂

  2. Lulz, all I heard was Facebook and Wal Mart.

    Facebook shifted toward a famous follow model in the last redesign with Pages becoming more like profiles. I never really cared for Pages.

    I suppose the enterprise will eventually need both friend/follow models, but it's still too soon to try IMO.

  3. Very interesting!
    Imposing symmetry maps a totally different social relationship.

    For example on essembly:
    * friends must confirm (symmetric relationship)
    * allies and nemeses don't have to (asymmetric rel)

    on facebook:
    * friends must confirm (symmetric relationship)
    * fans don't have to (asymmetric rel)

    I'm friend of May Bro (my brother), I'm fan of Barack Obama 😉

  4. We're experimenting with both models inside the firewall via Connect and OraTweet. I'm still not convinced that the asymmetric model will take off inside a company, but the recent influx of people on Twitter has me rethinking the value of that model altogether.

  5. We are developing and experimenting with Taolin in the research institute where i work (~400 people)

    But we don't have any definitive result for now! 😉

    I guess you already know but IBM research has done a lot of different work in the area: beehive (one platform, generated many papers, search “DiMicco”), but they are also testing other platforms internally, the ones they sell but also others.

  6. Not sure if we're talking about relationship models or reputation here 🙂

    We have the platform (Connect) and the data (two years' worth), so the next logical step is to focus on reporting and reputation.

    Re. relationship models, I'm skeptical that in a pure social network context like FB, that people in an enterprise would use the stalking model. I guess maybe after they get used to it on FB . . . dunno.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.