All posts made the week commencing Sunday 30 Mar., 2003:

Two favourite passages from Italo Calvino's Invisible Cities:

  • (From the introduction to section 9)
    And from the mixture of those two cities a third emerged, which might be called San Francisco and which spans the Golden Gate and the bay with long, light bridges and sends open trams climbing its steep streets, and which might blossom as capital of the Pacific a millennium hence, after the long siege of three hundred years that would lead the races of the yellow and the black and the red to fuse with the surviving descendants of the whites in an empire more vast than the Great Khan's.
  • (From Cities & The Dead 5)
    [...] you must think that the number of the unborn is far greater than the total of all the living and all the dead, and then in every pore of the stone there are invisible hordes, jammed on the funnel-sides as in the stands of a stadium, and since with each generation Laudomia's descendants are multipled, every funnel contains hundreds of other funnels each with millions of persons who are to be born, thrusting their necks out and opening their mouths to escape suffocation.

There's another extract on Purse Lip Square Jaw.

Thomas Kuhn's The Structure of Scientific Revolutions (at is excellent: how communities identify and understand a field through the use of paradigms, and how changes in these occur. A great mine of vocabulary for things I knew but was unable to articulate. And an interesting bridge between individual understanding, how people learn and communicate ideas, and group dynamics.

: Good synopsis and outline of the book.

More social software rambling:

  • Social software acknowledges that in the real world people like to work in groups of more than two, so this person-to-person versus person-to-unbounded_group in email is missing a step. (It's because there's an abstraction, or breakpoint at the sendee/receiver. The email address is just a pointer.) And therefore triads must be treated differently... but only as the first possible manifestation of few-to-few, which bleeds into other forms/axioms with larger groups.
  • Social software acknowledges that social capital starts to appear as people get together (favours swapped, debts owed), ditto groupthink, ditto norms/traditions: and therefore the design has to accomodate routes for people to reciprocate at the same proportionate effort-levels as real life, otherwise all members are going to feel annoyed. (More so: The software could complete the loop and make use of the concept of social capital too.)
  • An example of a feature that could've come from social software design: The IM-online icon in Apple's Mail. It would be better if it appeared closer to the email itself. Or maybe, closer to your focus of reading only when it's an email to/from a group, because then misunderstandings could be extinguished before they blossom.

And some rambling about social software as pre-paradigm:

  • Social software is like the early days of thermodynamics, before stat mech. Or maybe pre-Newton. Or maybe early electricity: we had to make do with rules-of-thumb. Some day all of this will be filled in with limits so we'll say: these approximations apply with this number of people; these others with this number. (This is already happening.)
  • It's different from this kind of science because the models only existed in maths, and now they exist in real life and people can live in them (Outlook). Or maybe not: having a good rule for electricity was important when they were dropping the telegraph cable across the Atlantic. Getting the amps wrong there was an expensive mistake.
  • Normal science/ exploration of an idea includes: fact finding, proposing theories, testing theories. We're proposing, but not testing. We're exploring, but not deriving the facts -- even if it's "this works, this doesn't"; we don't because with social software it comes for free: the ones that work are the ones that are successful. Successful and social should probably be decoupled so we can more easily examine them.

What we need:

Is a proper theory of folk psychology, a proper theory of folk physics, a proper theory of how people classify/understand. And not from the perspective of normal science, which is to say: "When these people have folk X, what is the real X they're thinking of?"

But rather: "What are the evolved attributes of folk X, so we can design for them?"

I like it when people say "I'm a tool guy". That means we (equals me. I'm a paradigm person myself) can take what they do, extract the attributes that made it successful, and reuse elsewhere. Some people can just create social software without thinking about it, like some people are great interior designers, or great orators, great at articulating themselves. Leaving these qualities in the hands of the people who were born with them isn't enough: that's why we teach people how to structure an argument, how to make use of rhetoric, why people go on courses for presentation skills ("What do I do with my hands?").

This is why we have Fung Shui: Folk interior design. Democratising aesthetic sofa positioning.

Social software is pre Fung Shui, pre folk anything. It's hard to slice and dice even, which is a characteristic of things that haven't been shaped and paradigmised. And it's hard because the term "social software" applies to all of:

  • the social software that people make
  • the process of understanding what it is about social software that makes it tick
  • thinking about social software and making rules, finding commonalities

And this is all without even defining what "social software" even means! We can't. It's like putting plumbing, engineering and fluid dynamics in the same box. But that's what electricity was like, what steam power was like early on, so there's hope.

No conclusion.

Arts and Crafts as a design aesthetic [thanks Matt Jones]: "Ideals: Simplicity, honesty (it is what it looks like), utility (form should follow function, but it also needs to feed the soul), organic, craftsmanship (includes innovation within convention), harmony (a simplified and unified environment). Used structural ornamentation - using joists and joints exposed. Wonders out loud how we could do that with web sites. Treat compositions of a web site as building on a series of harmonies". I like.

Slices of the ongoing conversation on social software.

Try and ground all of this; try and explain social software to someone who has a background in evolutionary psychology/ dynamics of group decisions, and it's very hard (as discovered a few nights ago). What's the objective? So I'll ramble:

There's no single defining feature of social software, no common thread. But some attributes which may or may not be shared: software which uses as data social relationships/properties; software which acts as an intermediary in social activity (conversation, decision making, wearing the same band's tshirt, clapping); software which uses human nature in the design process; software that has moved from providing an environment to providing an environment and tools, or more.

The human nature one interests me at the moment. Architecture or shop design does this: people walk into a shop and look to the right. Typography and logo design do: the total of the meaning of a logo is not just its form, but the form plus the perception. If the perception changes (over time, over a generation) then the form has to change in order that the logo stays the same. These are deliberate design decisions, but others are accidental/evolutionary: the physical form of cities, the operational form of social institutions (university, churches). These are shaped by feedback. And I mean evolutionary both in the short term (the evolution of successful social institutions; evolution within the institution itself) and in the long now: brain adaptations, real deep-down human characteristics. (Handshakes? Pointing? How the desks of powerful people are positioned? Cultural, or genetically emergent?)

Oh and of course! The www, instant messaging. Accidentally successful technologies. What makes them popular, useful, great? How do we build this stuff deliberately?

Then as for providing the environment. A group performing telic activity, do you give them email? Or something else? Email provides the environment and lets human nature do the rest. Outlook tries to give tools: meeting requests, calendars. There's a lot there. How has this been driven, decided?

So what are the problems to solve? How to make a technology popular? How to make people's lives more efficient/enjoyable? How to annoy the government into listening to you? How to work with someone on a document? But more than than: How to design software to defuse useless flamewars before they start.

Performance metrics: Two groups with the same task but different software. The group with this Software-Software-T.M.-enabled application will work faster, happier, with less fallout afterwards. And: Two technologies for sharing contacts, helping each other out. The Social Software Inside system will result in more favours, less bowling alone.

Look for rules of thumb. We've got some, for technologies: Look for the secret properties of software, copy them. We've got attributes and models (four types of players). I'm not sure these are enough.

There's so much to learn!

Look at philosophy, human behaviour, psychology, biology: extract commonalities of human nature. How do people work with distance? People prefer big effects to have big causes. How do people respond to pointing, clapping, shouting, unexpected people chipping in? Look at how people work in cities: those are the best examples of things-for-a-purpose with massive, group feedback loops. And how people sit round tables in pubs.

"You're looking to reconstruct the whole social world. That's a big job," was the response from the other night. Well yes and no. A cut down version -- but for that we have to know which are the important bits. We need a social rhetoric: a cut down way of acting socially, a way of democratising groups so that anybody (and not just people/software built that way) can be successful and happy.

I want a Fitt's Law for social software. Like: When there are twelve people who mostly haven't met talking in text, the chance of a groupwide flame war is 50%, so the button to respond person-to-person (over IM) needs to be a maximum of twice the cognitive distance as the button to respond to the group. For example.

So then: Experimental social software continue (FOAF springs to mind, weblogs too). Derive rules of thumb. People who are good at being theoretical, try and find paradigms. Learn about human nature - the kind of human nature which is important here, emotional response curves maybe - groups dynamics and so on. Model, break down. Put in the implicit design laws into the requirements document, things that aren't usually thought about: people need to be able to share their address books with this application because X, people need to taskswitch to this other app because Y. The purpose of this app is to do Z therefore there needs to be capability to 1, 2 and 3. And not just proximate: is socialising in group software just as important as the Meeting Request button?

Use cases, examples, models, falsification, experiment, theory, talking, above all: learning.

The promise of web services and OS-level scripting? Soybo is "a cross-platform and device independent technology that allows applications to publish their functionality as web services, accessible by any Internet-enabled device". Plug-in based and open, it seems to comprise a a collection of scripting hooks to expose applications and services, and b a collection of client applications to make use of those hooks in a pretty way. Incredible potential-filled screenshots. Kind of like Watson in reverse, or a high-level X Windows. [via Mac Net Journal.]