15may2002

Emergence


"Cities are decentralized in terms of the command system"
"Small pieces loosely joined" -- cluetrain/steven weinberger, so many people
have said this

general view: that many blogs will produce some superbrain.. not nec

cities good for:
clusters
public spaces/diversity
optimal density
-- and how does that apply to the www, using cities as a metaphor? that's what
this talk is about

--> spaces act as landmarks, and as mixing points
--> optimal density in neighbourhoods [so that's all about feedback loops in the
right places. that's a secret property of cities]
--> neighbourhoods organise themselves into communities
--> there is a 2 block neighbourhood for restaurant supplies

problem with "blogspace" is there is no neighbourhoods

[should we be building in the feedback loops to cluster, or should they be
inherrant in the system. we're not looking deep enough: if the system is complex
enough for these feedback loops to arise, they will. we need multiple distances
of interaction. what property of systems is it that allows feedback loops to
arise, given we don't know what feedback loops we have to write? or: do systems
gradually get more complex naturally until there's enough complexity for
feedback loops to exist? level 2 if you like]

[look. level 1 of the web is real + virtual. web + web is the recombinance, and
is natural in level 2. but there are some precursors already: search engines,
weblogs. anything that operates on the web itself. web services just make this
environment richer and able to evolve to the next level.]

he's trying to find the secret properties of the city
- bottom up interactions
- strollers [deciding to go a neighbourhood or not]
- passive organisations
- the swerve [being distracted by something walking from A to B]

he said "amazonian principle" for "people who bought this book also bought".

[interesting about strollers: like gauge bosons. like mediating particles of
interactions. carrying *something*, don't know what]

it works because of MANY swerves. over time. like google's scientific
measurement.

blogs:
problems forming higher level groups
overpopulation
tyranny of time
little passive organisation
readers lack inpuit
catastrophic success


[AHA
he's saying these are the secret properties that blogs need. maybe if the
environment is rich enough? still need to work on that.
LOOK
and
the tyranny of time thing is all about deep structure vs surface.]

he's saying you can't follow 500 weblogs

[man this is old
he's talking about running the blogposts through a clearinghouse to be
organised, clustered, syndicated
problem is the solutions are being built without the rich environment to create
the feedback loops]

he says need to hit "more is different" in the blogsphere

[what's the problem with blogs? no URIs]

pedestrian centric:
blogs come to you
search requests based on current documents
search limited to chosen bloggers + friends
variables

[this is rashid's query based thing as opposed to locations. query vs hierarchy]

so he's saying there needs to be a big database

[i think this is an incorrect conception. it's not BLOG centric, there's
positions in web space. generalise, generalise. each post is a point in the web.
concieve it as a path. navagate through it.]

[what he's recommending is big database. not going to work. ah, he's going to
hit this: the problem of standards]

but he's disagreeing -- that permalinks, datestamps, keywords and stuff CAN be
agreed on. hmmmmm. but self-organising keywords? ahh. he's talking about how to
generate the keywords by putting the smarts in the search engine. which is
interesting... but the flow needs to go both ways.

[maybe our environments aren't rich because the flow only goes one way, whereas
in the real world is goes BOTH ways. that's the problem is upstreaming... the
downstreaming relies on a person looking at both things. cory's nondeterministic
web thing]

[with google: good click/bad click emulates the swerve]

wants to see change over time in google searches. hm. so by looking at both
these types of searches you build up clusters of weblogs with both readers &
writers

he says:
blogs wont challenge NYT
blogs will challenge Google
[but surely google relies on blogs? or maybe he's saying that both things share
this web-on-web nature]

====

rumblings:
deep structure vs surface structure
new uis forcing a taxonomy
is anything nec better? eworld, filesystem, etc
www "secret property" is taxonomy can be redefined -> URIs
cities have secret properties too -- those things SJ is defining
query vs hierarchy is like the google onebox