Clay Shirky on Social Structure in Social Software


Clay Shirky is giving the final talk of the morning. Clay was the subject of some controversy during the conference caused by this article in the Register. Tim O'Reilly bit back saying He got a complaint from a speaker who didn't get included, and made that complaint the basis for a rant. He didn't talk to anyone at O'Reilly. He didn't make any effort to get background or hear the other side. He wrote a flame, not a story. I told Tim that I'm not surprised at all. Anyway, here's what he said...

Clay defines "social software" as software that supports group interaction. This is a radical concept given that most of the technology patterns prior to the Internet were point to point or broadcast. The best group interaction technology was the kitchen table. Clay's premise this morning is that social software is its own worst enemy.

W. R. Bion did research with groups of neurotics. He discovered that, as a group, the neurotics were taking actions that defeated efforts to help them. They were not acting individually, but there was not clear group consensus. Bion concluded that humans are fundamentally individuals and fundamentally parts of groups. Kind of like the wave-particle dynamic for electrons. Clay gives an example:

You're at a party and you're bored, but you don't leave. This is different than from a book store, etc. if you get bored -- you usually leave. That social stickiness is what Bion is speaking of. However, when one person leaves then everyone leaves. This triggering event (Gladwell's tipping point) pushes group over to action "Paradox of groups"

Bion identified some specific patterns of group interaction:

  1. sex talk done for the purpose of flirting or pairing.
  2. identification and vilification of external enemies
  3. religious veneration Nomination and worship of an icon ("cult of clay"). Putting something beyond critique.

Group rules like Robert's Rules of Order protect groups from falling into these patterns and thus protect the group from itself. Clay gives an example of BBS systems in the 1970's that started out as "open access" and "freedom of speech" were "overrun" by teenage boys who wanted to talk about bathroom jokes, sex, etc. The group didn't have enough structure to fend off these "attacks" on the group. This was a social issue, not a technological issue. "An attack from within is the pattern that matters."

Clay just said "they formed a government because they needed a government." I think this is an important point. Groups of all sizes need not only rules, but meta-rules (like a constitution). The probability that any unmoderated group will get into a flame war over whether to have a moderator approaches one as time increases. Small groups can engage in patterns of interaction that large groups can't. We are now getting interesting ways (RSS, chat, etc) to experiment with this, and ways to do the experiments quickly and cheaply. When you lower cost, interesting new things happen.

We're seeing an explosion of social software. Clay asks "why now?"

  1. Because it's now. Just learning patterns over time and internalizing these ideas. Clay asks why we had to go through Geocities with pictures of badly scanned cats before we got to weblogs, for example.
  2. Because it's web-native (weblogs, wiki's). He criticizes the "Lotus juggernaut with a Web front end". It's the web all the way in (not like lotus notes with a web plugin). Easy to extend, easy to break-down. Assumes http as transport, markup as encoding. This lets us build things quickly.
  3. We can now have a small pieces loosely joined (coupling, synchronous/asynchronous) pattern. One example is the emergent democracy work by Joi Ito. Joining of conf call / chat / wiki loosely linked into a mesh of modes. Interrupt messages move to chat room from conf call, and in chat or wiki, annotation happens. "Broadband Conference Call" held together with a little bit of social glue.
  4. Ubiquity. Everyone has access not just to the Internet, but also to the tools for writing the web. All is a very different amount than most. We no longer live in a two hula hoop world, where online and offline worlds are distinct. This quickly becomes an assumption.

There are some things that need to be understood to help make groups work:

  1. Do not separate technical and social issues. Not all social issues can be specified in technology.
  2. Members are different than users. In particular, there usually arises a core group of users in a long-lived group.
  3. The core group has rights that trump individual rights in some situations. Citizenship goes beyond just the ability to login. Otherwise you're subjected to the tyranny of the majority. All groups of any integrity have a constitution, whether in code or in social charter.

There are some design features:

  1. Identity matters. Clay says "handle" rather than identity because of the baggage that identity pulls in with it, but I think its important to deal with the messiness. Anonymous and even pseudo-anonymous systems don't cut it. The basis for group interaction is knowing who said what when. Moreover, reputation requires real identity. There should be a penalty for switching handles.
  2. Design some way for their to be members in good standing. There is an interesting parallel to a talk I heard by the Slashdot creators and how they came up with karma.
  3. Need some barriers to participation. This means that some things need to be hard for some people. This gives the core group tools to defend the group from attacks.
  4. You need some way to protect the group from scale. He makes a reference to Metcalfe's law.

Overall, regardless of what the Register article claims, Clay is a forward thinking guy and his talk was thought provoking.

Note on how this blog entry was created. A lot of folks here have started using a Rendezvous enabled editor called Hydra to do collaborative note taking. I've been logged into the feed, making some contributions, and freely using the group notes to remind me of workings, organization, and other things that happened during the talk, as it happened and I blogged it. Actually very cool.


Please leave comments using the Hypothes.is sidebar.

Last modified: Thu Oct 10 12:47:20 2019.