Will Streaming Media Stick Around?
ml at creativecommons.org
Wed Oct 29 22:19:31 EST 2003
Chris Karr wrote:
> Will Streaming Media Stick Around?
Small niche for real time distribution. If more than a few seconds of
delay is acceptable downloading will be preferred.
> Ok, so first of all, what are the advantages of streaming media over
> downloadable media:
> • The entire file does not have to be downloaded at once. You
> can start viewing the media as soon as you click the link.
That's just a matter of how a client is written. A file downloaded from
beginning to end can be played as soon as some of the beginning is
downloaded (no different than streaming, which requires some buffering
to give the user an unbroken experience), and indeed I've seen P2P
filesharing apps that can preview/play files while downloading. Even
with multi-source downloads/swarming distribution you can prefer the
earlier bits of a file so that playing can commence promptly.
> • Its easier to assert control over the use of the media.
Or, it's more of a pain to do anything with (including copy) streamed media.
> • The media is always up to date.
Except when the net or the streaming server is down or inaccessible.
> • Streaming media as opposed to downloading tends to not fill up
Disk capacity is growing much faster than bandwidth.
> From this perspective, there's currently a control and bandwidth
> argument that can be made for streaming over downloadable media.
I don't think there's a bandwidth argument in favor of streaming.
Streaming sucks over a slow link, downloading is slow. Streams require
bandwidth with every play. Unless streams are cached heavily on
clients, it will be harder to take advantage of client upload capacity
to expand virtual source bandwidth with swarming.
If streaming were a bandwidth win you'd expect <http://archive.org> to
> Using things like CC
> licensing metadata, the problem of checking if a file is up to date is
> solved by players that implement license metadata checking.
Similar technologies anyway. For example, RSS feeds of audio/video
content as a very simple beginning. Pushing the ideas behind the CC
embedded verification URL scheme (which just says there's metadata about
file x at uri y), you could easily imagine clients that looked for a
time range in dc:coverage and looked for updated content when the file
in hand is "expired". That's _without_ inventing any new elements.
IMO even where a streaming-like experience is desired downloading files
will be the preferred implementation.
You might be interested in the Alluvium (streaming experience, swarming
download implementation) presentation from CodeCon
More information about the cc-metadata