The Next Internet Infrastructure

Having registered and had breakfast, I’m in my first workshop of the day.

Jonathan Hare, Resilient
An open services archictecture. needs to be freely licenced, hostable, extendable and be capable of supporting a emergent ecosystem.
The web contains plenty of open content, but islands of authentication. Authentication needs to be first order in next generation architecture that is becomes possible to extend the way we do things on the web now to things that we currently can’t because of the lack of inbuilt trust mechanisms. i.e. you can’t apply web principles to healthcare, finance etc yet.
“openness is its own reward”, opportunity comes through interoperation. This doesn’t necessarily mean open source “google search is open, but not open source”

Jeff Barr, Amazon
Marc Canter points out freely licenced services have to be paid for - resources like bandwidth aren’t free.
Jeff talks about AWS, and micropayment model.
70% work required to build a web based service is “undifferentiated muck”, Amazon know how to do such things well and are franchising that out.
AWS aims to provide the things you can take for granted.

Chad Dickerson - Yahoo Developer Network
Yahoo Developer Network’s aim is to have open apis for all yahoo products.
As a developer, a great thing about open apis is that you can play around with services you potentially want to use before comitting to them. Chad talks about business 2.0, putting up open apis and allowing potential partners to play - apps build on these apis typically get built before any bus dev integration is done.
Jonathan mentions that apis are just really function calls, interfaces and anyone is free to provide alternative implementations This is certainly something we’re thinking about with regard to the Talis Platform services.
The real value is in the data collection plus the open apis, both are less valuable on their own.

Marc Canter
The path to developing new open standards for the web is to implement hardwired interfaces first, then try to drive standards through adoption, i.e. a leader in a space makes their interface stable and published (i.e. flickr) then adhoc agreement leads to stds.
Marc’s company, Broadband Mechanics is the developer of PeopleAggregator - authentication interconnection of the various standards for federated identity - providing single signon to accounts with proividers like YouTube, myspace etc using authentication archectures from Yahoo, OpenID & Cardspace.
Import & export via open standards like xfn & foaf
Very important is that APIS have to be 2 way - data out/data in.
Marc wants to enable sharing between data aggregators, i.e. for me to do a review in Yelp and send it to Amazon, very much thinking along the same lines as Talis.
Jonathan Hare: trust grows through time, requires identity to track behaviour through time & build reputation/trust.
Availability of anonymity is important - reveal the history, reputation stuff of identity without the actual identifiers.

Jonathan Hare: Expects personalised search + content delivery are some of the main kind of apps that could be build on an open identity architecture. The concept of anonymous indentity allows your history, behaivour and preferences to be shared to chosen partners.”most things that are interesting require a degree of privacy”

Random points from the workshop

  • Marc Canter: Caching is the answer to latency. Users of PeopleAggregator can configure caching per service.
  • Jonathan Hare: REST is important, if you follow those constraints/patterns it will scale inifiinitley. Easy to do for publishing (i.e. the primary domain of the web). No vocabulary for policy, authentication etc but the patterns of the web DO work for scaling. We’ve now reached the point where the web is pervasive enough that scale is the issue.
  • MC: “I don’t believe in centralised social networks”, the trick is interconnecting many disparate networks in an analog of the publishing web.

License

This work is published under a Creative Commons Attribution-Share Alike 2.5 License.