Last: Macromedia Central keynote
Next: From the margins of the writable web
April 24, 2003
clay shirky keynote
Clay Shirky keynote
looking for patterns of long-term large groups
social software - software that supports group interaction
Internet supports point-to-point, two-way, one-way outbound, and one-to-many.
Prior to internet, last technology that enabled group conversations was the table (conference calls are very hard).
Ridiculously easy group forming is new.
No particular piece of software is "social software". Email allows group conversations, but many other things other than group patterns. Weblogs are broadcast, but if weblogs talk about each other, then they're having a group conversation.
You can't specify in advance what a group can do.
Patterns of creating software, and being surprised when groups start interacting.
Part 1
Best explanation for this pattern is JR Byram(?) - doing work with groups of neurotics - together they conspired to quash their therapy. Nothing explicit, but it happened. Do they take action on their own, or are they working as a group? This is unresolvable. Do you view groups are a whole or as individuals? Has to be both. We're all intelligent, and have visceral emotions for one another.
Easy to see group cohesion when a group has a name - a guild etc. But this goes deeper and kicks in much quicker than expected.
You're at a party and you get bored - fails to meet a threshold of interest. You don't however leave. Social stickiness. When one person gets their coat, everyone does, because they're all thinking the same thing - but need a triggering affect. The paradox of groups - there are no groups without members, but also no members without a group.
Byram thinks this is the group protecting itself. There are specific patterns:
groups can devolve into
1. sex talk
group conceives of purpose as salacious talk between pairs of members.
2. identification and vilification of external parties
e.g. Linux in the mid-90s - mention MS.
They may not actually be an enemy, but this causes a pleasant sense of group cohesion.
3. religious veneration
e.g. try actually discussing Tolkein in a LOTR newsgroup
Group structure is necessary to defend the group from itself.
THis keeps it focussed on its own sophisticated goals, and stops it devolving to one of the above.
BBS called Communitree - open access, free dial up. Over time new patterns emerge. A high school got modems - the boys we interesting in fart jokes and sex talk. Communitree was overrun. "that's not the kind of free speech we meant".
The inability to defend themselves - technical or social problem? Doesn't matter. Tech and social issues are deeply intertwined. An attack from within is the pattern that matters. It was shut down by people logging in and posting. Split users into those that wanted it to continue, and those apathetic.
THis happens many many times, everywhere. "learning from experience is one up from remembering" - it's lousy. Not much learning from reading is all this.
assumed certain user behaviours - different to real behaviours. Couldn't decouple tech and social decisions. Lamdamoo wizards tried to only address technical issues - didn't work, had to go back and cover social issues as well.
People working on social software are closer to economists and political scientists rather than programmers. These social crisises are similar to constitutional crisises. Need to have rules for making some rules.
Jeff Cohen - chance of a group having a flame war about getting additional structure approaches 1 as time goes on.
Part 2 Why now?
Revolution in social software going on.
Internet was about how big you could be for last 6 years. Only works for broadcast, and no user conversations. Dense interconnected pattern isn't possible to keep alive on large scale.
mid-size groups (13-1000) are possible. Mailing lists, BBSes, IM, now weblogs, wikis, platform stuff - RSS, quick ways to build on top of infrastructure. - e.g. confab was built in two weeks. Cost is a lot lower now.
Because it's time.
Weblogs have nothing to do with technology - we got geocities instead of weblogs initially. Took a long time to realise people want to talk to each other (rather than scan pics of their cats). Pepys' Diary showed future - statement that blogs are going to be around for 10 years and more.
Now building things that are web-native - not Lotus-juggernaut with lightweight web interface. Not just on the surface - it assumes http etc.
Small pieces loosely joined - like emergent democracy. Same meeting happens in three different ways - conf call, chat (for signalling and annotation) [third?]
Social glue keeps them working together.
Ubiquity - most people have web access. And in some situations *all* have access. This is very different to most. Everyone you work with, your friends, your family are online. software now *assumes* that offline groups will have an online component - lightweight and easy to manage. Two hula-hoop world - one hoop of offline, one hoop for online. Now we're sweeping the hoops together. Face-to-face and online together at the same time.
All meetings now run with chat or wiki or something for those with wifi. Three weeks from never use it to assuming it'll be there [technology is indistinguishable from magic (til they get used to it)].
Part 3
What should we do?
Quit his day job because of Usenet in 95.
What makes a large, long group successful?
It depends.
Social software allows the same code to be run in several places - some work, most don't. Something supernatural (like supernatural anointment).
Universally true:
*have to* accept (happen anyway):
cannot completely separate technical and social issues
(Software determines what people do - up to a point - can't specify all social issues in technology. Group will assert rights somehow. Put responsibility into the group)
Members are different to users
(some group of users cares more than average about the success of the group. this is a core group. They care and garden the environment. Software doesn't always let the core group express itself - but they will find other ways to do it)
(members of good standing and drive by users)
Core group has rights that trump individual rights in some situations
(not one person one vote. It doesn't work when citizenship is as easy as logging in e.g. sufficiently contentious usenet groups aren't created. Wikipedia has volunteer fire department.)
Constitution - formal and informal. Formal is in code and rules.
Design for:
handles the user can invent (like identity). Weak pseudonimity doesn't work - need to be able to reference previous conversations. Best reputation system is in the brain [this is exactly the same as branding]. ebay works in atomic non-iterated transaction - not the same as social software. Has to be a penalty for changing handles - lose reputation or context.
Some way for members of good standing - posts appear with identity. Karma. Sponsors. (show member of good standing when you lose your sponsor's tag).
Barriers to participation - some cost to join or participate. Some kind of segmentation of capabilities - could be in or out, or partial, can/can't post, can/can't moderate. Pulls against ease-of-use. Ease-of-use is wrong - the user of social software is the group, not the individual. e.g. meetings where everyone has a good time but nothing gets done.
Spare the group from scale - scale alone stops conversations. Metcalfe's Law is a drag. Some way to hang on to "less is more".
Value is inverse to size of group - e.g. close friends, friends, acquaintances...
irc and mailing lists self moderate - gets worse, people leave etc. Don't make system large by taking individual conversation and blowing it up like a balloon.
Necessary but not sufficient platform to build on.
Explicit clustering, conversational artifacts (product is a result of process).
Hosting is like landlords and tenants - people behave as though they have rights. Users are there for one another - not for whoever paid for hosting and the software.
(see Tom Coates essay on social software)
cyberspace is not a separate realm - overlaps with real world. Support for communities online is social software - but these may support offline communities.
Q: why aren't we in 3d environments rather than text?
too small window, slow, like being drunk. Non-immersive software tends to have longer term effects - due to small pieces loosely joined. You don't always want to forget everything else going on.
Q: diff communities need different things - round tables, rectangular tables...
Current social software is architectural - have to go to some place. Haven't done well with goal-oriented software. Hard to make decisions in current software - but good for brainstorming. Add social layer e.g. on a wiki on how to use it. on-ramp, off-ramp pattern - do something and publish it - like Groove.
Don't have to design to be all things to all people.
Trackbacks
TrackBack URL for this entry:
http://www.anti-mega.com/cgi-bin/mt/addmttb.cgi/36
