2005-03-15

Rules for Remixing
http://conferences.oreillynet.com/cs/et2005/view/e_sess/6336


rael making a good distinction between customer and consumer, saying that
customer want to customise. good.

also remix and hack. hacking is remixing, but "remix" is inherently more
conversation. combinatorics. take, mix, push out.

#

tim quotes christopher alexander:

"each pattern is a three-part rule, which expresses a relation between a certain
context context, a problem, and a solution."

[compare with ramachandran's description of a good neurological explanation,
which we used in our hacks. from the proposal: "What: A functional description
of the hack, how it works, what evokes it, what parameters and constraints it
has. Where: Which neural systems are involved. Why: What evolutionary or
computational considerations have caused this". only interesting because i like
three-part categorisations at the moment.]

tim: "applications are no longer software artifacts: they're ongoing services"

[so it's not a delanda-like calcification of behaviour that creates
institutions, rather the stuff that gets spun off has agency. like improv.]

visualisation apps: http://babynamewizard.com/namevoyager/lnv0105.html

#

stewart butterfield on flickr shows "the 62 methods":
http://flickr.com/services/api/

there's stuff in the api that hasn't been exposed on the site yet because they
use it themselves.

[looking at the structure of the api names. stuff like "flickr.photosets.create"
and "flickr.tags.getListUserPopular". interesting. for evnt i was decoupling my
internal public methods and my imagined api, the latter existing in a flat
namespace. i shouldn't do that, evnt.event.create should be fine.]

#

firefox

future events:

- higher-level app events, instead of script-intercept and overlpa munging
	- this might replace overlays for some apps
	- events like "link-clicked", "page-loaded", "message-selected"
	- these are higher-level than a particular dom (eg, onLoad())
- "app sensors" to specify <command> enabled state

[this is fantastic--the OS needs these high level events too]

#

danny hillis, applied minds ("not a think tank, a make tank")

showing: videos of crazy types of model walking robots.

[they do big, insane projects for fun, and use them as bait to get real
companies involved to pay them to take it further.]

[talking now about a new kind of mass spectrometer that can look at all the
proteins in blood. this is useful because there are cancer treatments that only
work on 5% of patients, but nobody knows which 5. so you use biology and lots of
pattern matching and see what happens. this is good, looking for opportunities
all over the place.]

they do a lot of aesthetic work too.

one of the projects they're working on: being able to look at maps in groups
better. he made a map that can be spread out like a paper map. [it's a table
projection, and he's moving his hands over it to zoom into and rotate around a
globe. also drag the map. the table is also pressure-sensitive so he leans on
the table to move through different layers.

[great video]

this is a good project because there's software, hardware, and emotion, because
the map-makers fell in love with maps because they were shared and spread out
over a table.

another video is the next map table: a programmable surface. the table deforms
into the 3d shape of the landscape. [wow]

#

bezos, on a9.com:

adding a few elements to rss for paging and putting search results into this
rss: opensearch.

now at a9.com you can add more columsn: http://a9.com/-/search/moreColumns.jsp

this is really nice. you can add pubmed, and then do searches from a9.com into
those sites.

[two fantastic opportunities:
- aggregating the search results from things like pubmed, and making an
automatic, custom scholar.google.com
- having individual sites provide their local search results in this format.
then the sitemapdb plugin can grab those search results and move search to a
sensible place in the browser, falling back to a site specific search at google
if that's not available.]

anyone can submit columns to a9.com. you can search columns, and your bookmarks,
with the same search engine. [this has an absolutely tremendous amount of
potential.]

[i don't think rss is quite right. you need to use atom with a standard for the
uris, etc, in order that you can aggregate the results properly. semantically
structure the data.]