Re: DAP rechartering discussion

On Apr 1, 2011, at 9:42 AM, Alexander Adolf wrote:

On 2011-03-21, at 20:23 , Mark Watson wrote:

[...]
One thing that I like a lot about the way in which UC is currently specified is that one can easily split various bits off. There could be a UC Core specification with everything that's absolutely required, and then the other parts could be independent components. Since they're already optional and have a discovery mechanism, it should be relatively easy.

Fully second that.


In my experience from writing EPG apps though, I have to say that having some baseline search and metadata interoperability would be very, very nice. This doesn't have to be all-powerful (that's too hard a problem to solve) but at least some minimally common parts (navigating categories, simple full text search) helps a lot.

In the interest of provoking debate, I'd say that I'm not sure the concept of a universal "EPG" is valid in the "web&tv" world at all. There is no "Electronic Web Guide" for the web. We have search engines, but these are 'as thin as possible' to get you as fast as possible to the site you want: they do not constrain sites to be described in some particular metadata format, but there are tools for sites to describe themselves to search engines if they wish. If you know what site you want, you go can straight to that and, either way, the site then has control of your user experience when you get there.

Of course I could imagine why Mark would prefer a service-specific client. ;-)

So, there's an obvious reason and a not-so-obvious reason. The not-so-obvious one is because content discovery is really really hard. Search is actually a small part of the solution, because most of the time users do not know what they are looking for.

The people who have an economic incentive to solve this really hard problem are the service providers, not the TV manufacturers. A TV manufacturer is not going to do frequent software updates with new recommendation algorithms and better UIs - they already made their sale. The service providers have the incentive to continuously innovate to attract and retain customers.

So we have to imagine a world where service providers have the tools to be in a first-class position in the content discovery process on devices, not just the providers of the content once it is found. Otherwise we are stuck with dull EPGs for ever ...

And there should be nothing that prevented him from doing that or even would make it hard for him. It should be easy.

But I also agree with Robin that there should be a common baseline. Otherwise we won't be able to ensure a minimum level of user experience.

I'm not sure what this means. You mean that if you put the service providers in full control some of them might ship poor UIs ? Well then that provider will retain customers only as well as they deserve to ...

So IMHO there should be this baseline, and there should be pre-defined slots for proprietary extensions. To make them proprietary as opposed to closed, the extensions should be be identified in a MIME-type-like way. UPnP did the same for DVB metadata. They defined a generic container labeled "foreign metadata", and one of the container attributes is the "metadata format".

So, I'm not saying there should be no baseline metadata of cross-service search. But I think we should try to find ways that services themselves can participate more directly in any cross-service content discovery.

For example, rather than every service on the device providing metadata and the device performing a search on that, maybe the device should give the search term and any available context to each service and ask each one to return its results, then merge the results ? Maybe the display of each service's results should be delegated to the service ? The display of detailed metadata for a content item should certainly be delegated to that service.

Just thinking aloud, but what I mean to say is that there are better metaphors for device/service interaction than thinking of each service as a simple database of common-format metadata.


I think we should be enabling similar models for TV services and for multi-device services as well.

I'm nopt quite sure what Mark refers to here. Home network?


By multi-device services I mean example like searching on a tablet and having the content rendered on the TV. Or being able to interact with ads on the TV through a tablet. Interaction may or may not be directly between the devices over the home network.




Cheers,

 --alex