Re: Blogging needs a whole new BAG...

[Eric Scheid <eric.scheid@xxxxxxxxxxxxxxx>]

On 19/5/04 2:57 AM, "Bob Wyman" <bob@xxxxxxxx> wrote:

> Push vs. Pull: Currently, blogging is all about pull and polling
> Pinging: Blogging benefits from the various "pinging protocols"

There's a middle step which would be interesting to model ...

Blogs push a ping to the various clouds, which are picked up by the various
aggregator services, which then provide a feed of "these feeds have updated,
you may go check them now" (instead of acting simply as a proxy service).

Thus: many++ subscribers regularly poll for just *one* feed from some
service, the service may or may not pro-actively poll (relying on pings)
making just *one* regular poll to any given blog, and then when updates
occur the subscribers pull from the blog (thus alleviating the service of
burdensome bandwidth).

For this to work 
(a) the desktop aggregators need a bit more smarts in their
polling/scheduling code (some only have a global period setting!),
(b) the feed format needs to support references to feeds as entry items
(concept proof already done)

When I say "interesting to model" I'm referring to figuring out the real
numbers involved, which would require knowing how many feeds people
subscribe to, how much overlap there is, and so on.

combine that with the (S)SFF method of granularity, and I expect some pretty
good bandwidth savings. Heck, if we can shift this from O(n) to O(log n)
that's a win.

e.