[mythtv-users] possible p2p approach for mythtv information?

[tom roth]

> Sounds to me like you are creating "networks of interest" or "interest
> groups" since not everyone has the same viewing
> preferences and people with similar viewing interests would naturally
> reinforce each others recommendations. The
> basic interest groups would be those that receive the same service, i.e.
> the same broadcasts (since it does not make
> any sense to take a recommendation for a recording that is not even being
> broadcast in my area).
>

Exactly, i think the grabber, that is used could be a good  identifier.

For arguments sake I'm assuming that my own recording selections
> automatically become recommendations for others,
>

right

as a default setting.
>

Wheter default or optionaly is open to discussion i think. Many might
dislike the idea of mythtv sending data into the internet.

To add some spice, it would be good for me to receive feedback, i.e. if my
> recommendation is
> not accepted by other peers then I should be notified of this, and
> similarly if it is archived. This would allow me to not
> broadcast recommendations that are known to be unacceptable, thus reducing
> traffic (i.e. rather than be dropped at
> the other end).
>

on one hand this of course depends on wheter all recording selections
automatically become recommendations or wheter you explicitly post a
recommendation. I would tend to have this system as unobstrusive as
possible. Also anonymity is an issue to consider. It should not be possible
to spy on a single viewers preferences.

You'd need sensible ways of dealing with conflicts (i.e. there are limited
> inputs available and like many recommendations
> that cover many more channels than available inputs) and with spam
> recommendations. Negative reinforcement can
> eliminate a given IP address but spam will likely use fake IP addresses
> that could flood the incoming recommendations
> with faulty or biased data.
>
I guess you are suggesting a ranking scheme that admits high ranking
> recommendations over
> low ones. But then this makes it harder for newer and odd recommendations
> to enter  the system. There is a trade off
> that would be interesting to explore.
>

Yes, this is an issue. But my experince with gnus as a newsreader were
pretty good. I ended up with the news highlighted that i was actually
interested in.
For different newsgroup i was also able to define different tresholds,
depending on the volume off the list and how much time i spent on each list.
So i could have a low treshold for documetaries and a higher treshold for
action movies for example. I would end up haveing more documentaries
allthough the volume of action movies would be higher. (thinking of scifi
would probably rank pretty high)
Also i would not expect the system to work flawlessly , i would  just like
to come home and find mythtv has recorded something for me allthough i was
to lazy to  do my programming ;-)
I guess it would beat live TV for me most of the time.

I like it. What approach would the interconnectivity/peer discovery take?
> Gnutella? Napster? something else?
>

Since i was inspired by gnus, a nntp server comes to mind. A recommendation
would tranlate in to a posting that is then filtered by the client. Existing
nntp libaries could be used.  Not sure  this is the best approach for
scalability. There are so many interesting protocols out there .

-a
>
> PS>I'm sure that the broadcasters would love to know what people are
> recording! No doubt they would have a peer in
> the system as well, monitoring what is being recorded! (or at least, what
> people are recommending to record)
>

Not sure wheter the mythtv crowd is reprensentative for the mainstream;-)
Maybe this data could even be traded for good programmdata access.


rawdlite
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20060703/4cd09a67/attachment.htm