Shared Motion : Follow up

Dear IG members

Thank you all for your interest and assistance in identifying potential
relevance to W3C. Special thanks to Mark and Guiseppe for helpful
suggestion on how to proceed. See below for our comments.

As noted earlier we’d like to share a paper presenting the full argument
behind Shared Motion (see attachment)

Note: the paper is styled as a scientific paper - though it has not been
peer-reviewed or published anywhere else. We plan to make a proper
publication based on this later on, so feel free to share some comments.

Here’s a brief outline of the context of our proposal.

As very generously indicated by Paul Gausman (this list, 16.des 2014),
“distributed temporal control” has important implications for HTML5, for
the user experiences it can support as well as the systems that support
those user experiences.

Particularly, in the context of broadcasting, distributed temporal controls
address a wide array of issues and challenges;

   -

   time-consistent presentation of live web-content from multiple sources
   -

   synchronization of web-content with live broadcasts as well as on-demand
   content
   -

   switching between devices without losing the temporal context
   -

   sharing temporal context with other people
   -

   time-shifting of web content to match broadcast distribution delays
   -

   collaborative viewing or authoring along common timelines
   -

   time-sensitive recording of user input and crowd-sourced content
   -

   monitoring of temporal navigation of viewers
   -

   time-sensitive advertisement
   -

   and more


To our knowledge, Shared Motion is the first mechanism proposed for
distributed temporal controls that meets central requirements of the Web
domain; simple usage, reliability, availability, precision, light-weight,
scalability, general-purpose, quick load, client-server architecture, etc.

We have been doing Web development using Shared Motion since 2010. Based on
this experience, we have identified sources of friction that keep Web
developers from leveraging the full value of this mechanism.

Our proposal then targets removal of those frictions. In particular, we
point to specific modification to the HTMLMediaController, HTMLMediaElement
and HTMLTrackElement, that, if implemented, would significantly improve
HTML5 as a platform for timing-sensitive multi-device applications.

So, our main proposal is not really concerned with the protocols supporting
Shared Motion, but with existing concepts in the HTML5 spec. (I apologise
for being unclear on this point before).

Concerning the first exchange with Guiseppe Pascale (15 des), we certainly
agree that Shared Motion does not require standardisation as protocol in
order to be available for Web developers. However, I personally quite like
the idea that the Web should include a mechanism for distributed temporal
control. This could even shift the way we perceive the Web from a
distributed system of hyperlinked documents, towards a distributed,
time-sensitive, hyperlinked rendering machine. Now, with Shared Motion as a
specific proposal for mechanism, there can at least be a debate.

And by the way. The protocols used to implement Shared Motion are strictly
client-server and based on existing Web technology, i.e.JSON over WebSocket
or HTTP. (I probably should have mentioned this earlier as well)


How to proceed

Reporting some of the issues with existing HTML elements in regards to
synchronisation as bugs sounds like a sensible idea, but it should likely
be backed by interested parties in order to motivate a serious discussion.
For potential standardisation of Shared Motion as an interoperable way to
handle distributed temporal control, a CG is likely a better option. There
already exists several existing groups with some relevance, e.g. Secondary
screen presentation, TV control, Timed Text, Web MIDI and Web Animation.
Secondary screen presentation and TV control, address distributed issues,
but adopt a very different approach. TimedText and WebAnimation are single
device, and Web MIDI is very tied to the MIDI protocol.

We think it would be a good idea to have a CG dedicated to distributed
temporal control. “Web-based Multi-device Timing Group” or something like
that. It would be great if such a group could grow out of the WebTVIG.

We also like Guiseppe’s suggestion to focus discussion on established
scenarios. We’ll follow up later with some thoughts and demonstrations on
how Shared Motion applies there.


Best,

Ingar Arntzen and Njål Borch