we did http://en.wikipedia.org/wiki/Six_hats over the topic:
white hat (facts, figures, info)
theres lots of data
federations connect multiple independent data sources / many nodes that act as one (much like the BORG)
there is a difference between raw data and URLs
there are lots of formats
lots of different licenses
federation is many nodes that act as one
person generating data could be responsible for it
estimating sychronicity is hard
there are different metadata formats (stuff)
google is indexing this stuff
esri bought the dream
there are huge efforts on GIS side to do this
libraries are a similar model
GeoRSS is a federated format
FORMATS
GeoRSS, KML, Microformats, Atom, GML, shapefiles, .osm, OGR, WKT(B), Raster :-( GTFS, GeoJSON, HTML, CSV, TSV, Excel, GPX (standards are great there are lots of them)
red hat
federated is more valuable than centralised
Google might be evil
Scared of Google not doing what we want to do
Google changed definition of open - transit spec - urbanmapping guy. ian white - who owns the data?
Federation might just be too hard to achieve
There is no One True Format
KML is horrible
Needs to be format agnostic
People are standardising on REST
There are other formats - REST might FAIL
how the fuck do we pay the rent? - biz models
Are we locked in to the advertising model
Licenses are going to stop much of the cool stuff happening
YELLOW HAT (positive happy thoughts)
Google are not evil
GeoWeb is super cool / has a lot of momentum
Federation is a nice dream
There are viable biz models
People are supporting standard formats (mikel keeps hitting them on the head)
OGC is taking notice IFTF, W3C etc
There are lots of success stories / tools - mapufacture, UN things... the internet / google maps / Earth
People want this stuff
Build it and they will come
It's getting cheaper / (easier?)
People need to read the Hirsch report on peak oil (DoE report)
Federating creates meaning
SemWeb++
There is more and more data (say, oil deposit data)
In the future there will be even more Geodata
BLACK HAT
This stuff is fucking hard
Security
Not much international data :-(
Licenses suck :-(
Govts are super unhelpful on this
Licenses might not even work (See the Open Database License ODbL)
i18n is hard
Trust models are hard / dont mach real world / disfunctional as ths UN
GeoSPAM
Conflicting datasets - who to trust?
There's too much information
Privacy of location might be hard
Centralising this stuff is hard (unless you're google)
some People dont like google centralising things some of the time. sometimes.
Our fun and games can be used to blow up iraqis
Timing is going to be hard... data gets out of date fast
The big formats are not necessarily useful (KML?)
GREEN HAT
Maybe we should look out of the box
GeoTorrents!
maybe we could be format agnostic
Realtime models... not thinking about caching and stuff
feedsync? ray ozzie dev.live.com/feedsync - federated feeds and stuff
XMPP pubsub spec atom over jabber
Yahoo! pipes for maps
feedblender - gone!
GIS industry have been debating this for 30 years - maybe we will be for the next 30
JFDI ?
release early, release often - lets JFDI
this stuff needs to be up to date - is the chinese resteraunt still there / open?
Maybe there should be more than one way - realtime vs bulk
build something that works to demonstrate value of federation to get more data onboard (examples: http://code.google.com/p/googletransitdatafeed/wiki/PublicFeeds)
Comments (0)
You don't have permission to comment on this page.