The always-interesting Jeff Jonas wrote recently about outbound record-level accountability, i.e., tracking where sensitive data is sent.
Without outbound record-level accountability … ensuring data currency across information sharing ecosystems can be problematic. The challenge being when a record changes in the originating system, how will one be certain which recipients of the original record need to be notified?
He adds that while such accountability is desirable, “not every mission will warrant the cost.”
I wonder, though, how much of the cost might evaporate if we make the architectural shift from sending data around, like email, to publishing it, like blogging.
I love the phrase data blogging, which Gavin Carr coined in response to some of the articles in my hosted lifebits series. One of the things that falls out naturally, in a syndication-oriented architecture, is the ability to audit who your subscribers are, and which chunks of data they access.
Note also that Jeff’s caveat about “which recipients of the original record need to be notified” implies owner-initiated push. But if the recipient is a subscriber, that update channel is already open and ready for use.
In terms of the value that the syndication pattern can provide, both for inter-personal as well as for cross-organizational communication, I think we’ve hardly scratched the surface.
2 thoughts on “Syndication and accountability”
Amen! It would save me a lot of heartache at work if I could get the medical records people to understand this.
“Note also that Jeff’s caveat about “which recipients of the original record need to be notified” implies owner-initiated push. But if the recipient is a subscriber, that update channel is already open and ready for use.”
Hmm. We need to be clear on the division of responsibilities here: The owner is responsible for making notifications available, but the client is still responsible for initiating the retrieval.
I think Dave Winer observed once that most ‘push’ architectures are actually ‘scheduled pull’, and that pretty much all successful architectures are pull-oriented, at least at the receiving edge.
‘Push’ only seems to work from the edge to the center (if there is one), or from the edge to the cloud (if there isn’t).