Device Coordination 

What is device coordination?

Everyone has their own view.

One view is accessing a device from web scripts.

It could also be in something else e.g. markup.

What about security concerns, browsers operate in a sandbox.

Answer - browsers need a policy-based mechanism.

Aside on bluetooth virus to someone walking through an airport
and was pestered until he said yes to the bluetooth bonding
request.

Device coordination may have a broader meaning to some people.

Principally messaging between devices as a means to coordinate
activities.

Does the discovery mechanism need to understand how the
discovered entity fits into a workflow?

Are really talking about devices and what are devices anyway?

Surely, we are more interested in services?

Dave suggests that discovery is about resources as a more
general concept above devices, services and data.

You may want to discover a capability that is only satisfied
by an aggregation of services.

Would you go for searching for an aggregator or a planner?

Johan notes that printers already aggregate a range of
services.

But what about behavior that is emergent from an aggregation
that is assembled upon demand.

Franklin likes the idea of a planner that figures out what
aggregation needs to done.

Example of transcoding PPT to PDF via a web service that you
can search for.

Johan notes that some devices can't support a planner on
account of limited resources.

Dave responds that it can be delegated to HTTP servers
and can be anywhere, even on another continent.

Caching may be useful for perfomance reasons, e.g. keeping
the top ten solutions.

Anything that can reduce the planning space helps.

Planning has been worked on for many many years.

What hasn't been worked on is how the requirements to
be addressed by the discovery is done.

Ad hoc discovery works well on the web today thanks to
the vast amount of content out there and the effectiveness
of search engines.

Franklin proposes a working definition of device coordination
as:

There needs to be some set of representations of information
about devices and services, and secondly, there needs to be
some means to use this information to create a solution to
the request.

There may be commercial pressures to direct the search in
ways that benefit particular businesses.

Karl notes that users may indicate preferences in how
requests are satisified in particular contexts, e.g.
when walking around a city.

Johan reckons that a reputation based mechanism for
figuring who to ask requests would be valuable.

Don notes that location information is a valuable piece
of context.

Dave says that this is something that has been discussed
a lot in W3C work on DCI, with the use of 3rd party
services to transform location information and the need
to keep things simple for application developers.

Registering your preferences with websites is also
important.

There is huge market in providing information about
visitors to websites especially ecommerce sites.

Different applications have different requirements
for location e.g. different precision.

Karl notes that you may be willing to provide context
data to sites according to a set of policy rules.
This would reduce the need to repeated registation
of personal information.

Johan talks about robots which unlike humans have
very little understanding of the context.

Dave walks through his smart meeting use case which
involves combing multiple sources of information
about employees, offices, calendars and meeting rooms.

Franklin agrees that that could work, but feels that
in other more informal settings, guessing goes wrong
too often to be worthwhile.

Dave reckons that given enough people they will find
some novel and yet valuable applications. This needs
a basic framework in which applications can leverage.

Context awareness doesn't mean you have to give
everything up.

It is important to pick a good set of use cases for
the work on developing a device coordination to succeed.

For humans, lying and cheating is a feature not a bug, 
says Karl.

Intentially misrepresenting the truth is ...