Response to comments inline on final report

(@@Mischa: The notion of "portability doesn't mean anything to me, for
me it implies a walled garden or a data-silo@@ @@Henry: I agree that
this notion is overemphasized. Machine readability and linkeability
are more important @@)

Yes, but for many people dataportability may mean export of their data
back in vcard, which is machine-readable. And many sites are de-facto
walled-gardens. Given that this word is widely understood and "machine
readability" and "linkability" aren't, I say we mention the previous
two terms but keep "portability."

(@@Mischa, perhaps this is a good point in time to talk about the
DataLiberation Front at Google, which falls under dataportability)

Agreed, will add in.

(@@Mischa, there have been a number of conflict reports in this space,
I have a write up of them here :
http://mmt.me.uk/docs/dsnsChildrenPrivacy.pdf, and would like this
expanded upon)

OK, will try add in.

@@Henry Story: The notion of context though important is
overemphasized in this section. Either the section should be renamed
to "Social Web Context awareness" or the content should be changed.
Context is important, but it is ill defined here. @@

Not sure if "awareness" is a better term than context.

(@@Mischa, I am not sure how "time" comes into this, perhaps it could
be elaborated upon@@)

OK, meant "latency" I think

(@@Mischa : I don't think that privacy and security are the same
thing, and don't think they should be confused))

Agreed - let's define those terms in bold. Any good definitions?

(@@Mischa : I would like to see the group describe how the
architecture, that is the web, can cater for Social Networking
functionality without the requirement for a third-party service
provider to store, own, and have access to all of your personal
data@@)

(@@Henry Story: I agree this whole section is much too much tied to
the notion of third party providers hosting a site @@)

Agreed and will try to edit that point in.

(@@Mischa: I am not sure what this forth columns in meant to illustrate@)

Mapping to other lexicons.

(@@Mischa does a Social Graph contain the graph of all "likes"?@)

Yes, and good example.

@@Henry Story: That does not put the emphasis quite in the right
place. The distributed Social Graph is a set of relationships between
agents whose profiles are on different domains. Of course in order for
these different domains to read the information from the other domain
they must communicate with formats with clearly specified semantics.
Just as Facebook "Like" Button allows you to like pages on other
domains, so a distributed Social Graph, should allow you to "like" ie,
friend, people on other social networks -- I don't find the icon very
easy to understand.@@

Correct, will try to edit that in.
*

@@Henry Story: It would be more useful to provide links to sioc and
foaf vocabularies@@

OK, will try.

Distributed Profile

@@Henry Story: The above does not show how the profile is distributed
or how it can be merged. The picture below uses the widely known
database symbol, a cyclinder - Oracle's campus near Redwood city is
composed of 5 cylindrical buildings! - to show the information
contained in the database. The two small graphs show the information
fetched form social networks - it would of course be better if in a
previous diagram those social networks were in fact shown to be
isomorphic to this. The large graph shows the merged information. @@

I don't care about graphics and will not create new ones. If someone
can edit then to do better, go for it and I'll edit them in.


@@Mischa: There are identity providers out there (like the
SocialAggregator described above), I think they tend to be called
"Identity Selector"s. There are a number of Identity Selectors out
there in the wild which allow you to select which "infocard" profile
you wish to identity yourself with to a given service, these include
and are not limited to : Windows CardSpace, Bandit Project's
DigitalMe, and Eclipse's Higgins Project has something in this space
too@@

Agreed, will mention.

@@Mischa = The phrase "locally on her browser" is confusing to me,
"locally on her machine", makes more sense@@

"local on device"

@@Mischa -- in email, only server to server comm "tend" to be
unencrypted, but *i think* there is a facility to do encrypted server
to server comms@@

Yes, StartTLS. Will mention.

@@Mischa: Again, why "in the browser", I mean even if i am interaction
with http://evildude.example/ I will interacting via my browser. This
distinction seems odd to me.@@

See Firefox Sync/Privacy icon work. It's where it can be implemented
locally most easily.


@@Mischa: This is the first bit in the whole of the "Business
considerations" section, which is actually applicable to the world of
a "decentralised social network" what I believe to be the only aspect
of the "Social Web", which is actually novel, empowering, and worth
working towards.@@

Agreed. Will try to add in.

Rest of comments seem to be stubs, will try to fill those in.

Thinking of removing use-case appendix to separate doc.

     cheers,
         harry

Received on Wednesday, 15 September 2010 14:59:29 UTC