[FoRK] PubSub 2.0

B.K. DeLong <bkdelong at pobox.com> on Mon May 7 14:50:41 PDT 2007

has nobody paid attention to the power of social networks for event/spam

This is where trust-relationships and customization come into play. My
thought is to always give the user the control - give them the option to:

1) see everything automatically
2) See automatically only what's available based on their (hopefully)
carefully crafted profile / interests / whitelist / blacklist
3) Turn everything off and at their leisure, see what's available or
everything based on profile specifics.

Ideally PubSub or whatever would have a built-in social network component
that would have a set of standard relationships (family, business, friends,
acquaintances), and subrelationships (spouse, sibling, manager, subordinate,
good friend) and then a some sort of means for designating trust levels in
that particular person or relationship type/subtype.

To be further ideal, people would be able to create their own relationship
"groups" or filters ala Livejournal filters and assign scoring to them.

In a perfect world, Software Agents would learn and make "best guesses"
habit-wise (where's my virtual butler?!), but I think this combined with
Amazon.com and TIVO - like preference learning combine with shared
blacklist/whitelist scoring of trusted relationships, we can sure go a long

Usual "Raw Thought" Disclaimer applies.....

On 5/7/07, Jim Whitehead <ejw at soe.ucsc.edu> wrote:
> Two thoughts:
> * Isn't this PubSub 3.0 (or 4.0)?
> * So, I walk out of my house, turn on my cell phone, and I'm
> immediately hit with event-driven porn spam?
> - Jim
> On May 6, 2007, at 1:53 PM, J. Andrew Rogers wrote:
> >
> > On May 6, 2007, at 1:07 PM, Ian Andrew Bell wrote:
> >
> >> I'm confused.  I don't understand how (below) is related to
> >> PubSub, so there must be some aspect of the company's goals that
> >> my postmortem forensic analysis might be missing.
> >>
> >> I have noticed that the company is getting raw Tsunami alerting
> >> data from the USGS <ens at usgs.gov>.   I assume that they had built
> >> some tools of some sort to cluster/aggregate news and reporting
> >> data after the Indonesian Tsunami.
> >
> >
> > Sorry, it kind of went off on an unrelated thread.
> >
> > After the tsunamis the governments got hot for subscription-driven
> > event notification in a geospatial context, primarily for disasters
> > but now they want to take it much further and apply such a
> > framework to all manner of thing that may have nothing to do with
> > disasters per se.  Child abductions are among the very first things
> > they want to do this with, for example.  Fox News will be all over
> > this (*rolls eyes*).  The basic model looks like this:
> >
> > The individual subscribes to any number of specific event feeds
> > with a geospatial context, voluntarily or involuntarily depending
> > on the feed and the government.  Some of these feeds are for very
> > specific types of events, other ones are descriptive and abstract.
> > As the individual travels around, the events they see are the
> > intersection of their subscription profile with the event, and
> > their geospatial location with the geospatial area of the event as
> > defined in the feed.  The problem is scaling up the ability to
> > determine if an individuals subscription and event intersect in
> > near real-time, and this needs to have a significant amount of
> > centralization as a practical matter and political control along
> > the usual boundaries is a big issue.  Both the events *and* the
> > subscriptions will be floating around in space constantly, and this
> > needs to be sorted out if they are going to use it for anything
> > remotely resembling all they want to use it for.  There is a full
> > expectation that the private sector will rapidly find uses for it
> > too, for good or ill, piggybacking on the government mandated
> > interfaces.
> >
> > Do not worry about signing up for this service, it will be
> > mandatorily built in to your cell service at a minimum.
> >
> > Cheers,
> >
> > J. Andrew Rogers
> >
> > _______________________________________________
> > FoRK mailing list
> > http://xent.com/mailman/listinfo/fork
> _______________________________________________
> FoRK mailing list
> http://xent.com/mailman/listinfo/fork

B.K. DeLong (K3GRN)
bkdelong at pobox.com

http://www.wkdelong.org                    Son.
http://www.ianetsec.com                    Work.
http://www.bostonredcross.org             Volunteer.
http://www.carolingia.eastkingdom.org   Service.
http://bkdelong.livejournal.com             Play.

PGP Fingerprint:
38D4 D4D4 5819 8667 DFD5  A62D AF61 15FF 297D 67FE


More information about the FoRK mailing list