Architecting – A Self-Regulating Society Theory and Practice

Session Topic: Architecting a Self Regulating Society

Thursday 1F

Convener: Matthew Schutte

Notes-taker(s): Sarah Allen

models & methods of a self-regulating society?

methods for enabling emergence?

matthewjosef@gmail.com

15 years of thinking about how you enable government at global scale
 * need to get the lowest levels right, not too much responsibility, enable innovation
 * what are those components?

Question: what do you want the outcome to be?

wrong way to ask the question: change the bottom levels, and see the system that emerges

actually: platform can be general

humans and all systems regulate themselves in some ways, humans have used technologies over time to enable collaborate

language, writing, print, the internet

belief: if you push power to individual decision-makers, not only allow them to turn to or subscribe to resources

also control what they put out

if you build this stuff in (privacy, reputation), it will be a self-healing system with opportunity for redemption

what is the content?

how do we create graphs of relationships?

not requirement of forcing agreement of the identity of things -- I make a cell phone and you make a cell phone, but maybe they aren't the same thing... you could make a map of the similarities and difference of the things, allow people to understand how to navigate between them

this is how ants work... if an ant has done the work of exploration, there is some signaling

3 principles (someone notices from initial summary)
 * continuity over time
 * reachability
 * designations that are persistent

what about forgotten? story about blah blah car, conversation about public reputation -- allow sharing to happen, but check for time

you aren't going to be able to solve this all with technology, you also need etiquette

and...

Bruce Schneider book -- as societies we enforce behavior, mechanisms are different at different scale

hypothesis: if we build this mechanism well, then the problem is simplified, even at large scale, we can respond effectively to social pressure

don't expect this to replace government, but reduce the need for police enforcement

XDI. RDF enabling you to navigate relationships

content addressability

feedback mechanisms

reputation of an object -- need to have meta-data

voluntary dissociation -- freedom to leave without force keeping you there, may be

social repercussions

attribution -- who did what (authorship may be too strong a word)

data lineage

Let's get down to the lowest level:

what does an ant need to talk to another ant?

Signals?

signals that people send back and forth to each other, have rate & context

ants have simple rules they are following

who are the players?
 * autonomous agents or people or ...?

what mechanisms do you need to start from disjoint nodes and build a social graph?

thinking back to the beginning of time? what do people need for collaboration?

trust & ethics

response: push that to the users

discussion of crime, negative & positive reputation, with systems of contract,

insurance, and reputation systems (escrow is an important part of this)

is crime relevant to the purpose of collaboration? if someone creates a new identity & creates a bit of knowledge, do I care that they committed a crime with their real identity?

immutable information, you don't necessarily delete things, but rather add markers that it is no longer relevant after a certain point in time

graph theory

how to address things? what about a URL? URL has domain name in it, which assumes it is remote (or relative which is imprecise)

we need some unique identifier, use git as a model

note: Zooko's triangle

low level mechanism : reliable designation

no one naming system can all have three

memorability is something that content addressing doesn't have

the HCI of content addressing is not that you look at the address and make sense of it -- you look at the content and know what it is without human readable addresses, the web would actually be less breakable

this is part of Time Berners Lee's initial vision of the semantic web, but he couldn't build Xanadu

there is a level of maturity of supporting technologies that enable us to do new things

not enough to understand that something that went before didn't work

not that the vision was wrong

important to know why it didn't work

Granovetter designation system, "pet names"

It is significant that we can find something, not what it's underlying technical content address is

anti-requirement -- it is ok if 2 people (or two things) cannot talk directly

two people could have different graphs of meaning that refer to a single object, objects have annotations which are user's meaning, definition of what they are

Objects can
 * able to designate itself to others
 * able to pass on other's designators
 * specify actions
 * permissions -- if you don't have permission to use it, then it should act like it doesn't exist

UMA (User Managed Access) & OAUTH might be useful (as concepts), but details will need to be changed since these are http based