[foaf-protocols] Important Video from Zeitgeist 2010 re. Linked Data and Privacy

Nathan nathan at webr3.org
Thu May 20 11:11:26 CEST 2010


Kingsley Idehen wrote:
> All,
> 
> Do watch and digest: http://www.youtube.com/watch?v=10VXT0aGICc .
> 
> All roads lead to fixing the Identity problem.
> 

Very good video, thanks for sharing Kingsley :)

I was rather sad to see that there seemed to be some conflation between 
"linked data" and "linked open data" where important points such as this 
data is already open, it's just being made available in a machine 
readable format, were missed. The decision about what should be open and 
what shouldn't is somewhat different to the discussion of linked data 
and universal data models & formats.

No point was made that linked data can be adopted behind the firewall, 
where data is private, and yet can still be augmented with the wealth of 
public open linked data that is already on the web. I personally feel 
such an example would have made a clear distinction between the issues 
of open formats / linked data - and which data is actually 'open' for 
public consumption.

Moving back to context though, it perhaps did hit home the importance of 
things recently discussed but not yet covered in depth or finalised; 
such as the splitting of data about a subject over different documents 
and graphs; encrypting or acl controlling that data.

For instance I'd like to have my family name, private email address, 
cell phone etc all available as linked data, in my foaf, but not public 
- my own personal preference is that this wouldn't even be open in an 
encrypted format (technical issues aside), I'd much rather have it in a 
partitioned document to which the people I foaf:knows were privileged, 
and on which I may one day monitor access and perhaps revoke certain 
permissions if I note the data is being accessed by somebody I 
foaf:knows but using a tool I know to be malicious.

Also it smacked home the importance of inverse functional properties 
such as mail sha1 & md5s, things we may not consider to be personally 
identifiable, yet as we all know are, they can be used to link the data, 
even in contexts such as sioc:UserAccount where there it's not marked as 
an inverse functional property, it can be used across the giant global 
graph to at least say "person x holds account y on service z" that alone 
could be data people want to keep private & control.

Additionally it's worth remembering that many properties when combined 
can be used to identify and link entities, a simple example would be 
that if there something with a label of webr3, talking about the 
subjects covered here it's likely to be me - owl:equivalentClass 
combined with owl:Restriction is a very very powerful tool for this, 
whether working with "links" or literals.

As illustrated by the video, (even if not mentioned directly), the key 
to many of the issues they mention is putting people in control of their 
own data; and as we know foaf+ssl combined with the notion of personal 
access controlled writeable data spaces is a very important milestone to 
reach.

Indirectly, it also brought to my attention the importance of a single 
channel of communication, or should I say a funnel of communication 
which we can control - the web at present has multiple disparate methods 
of communication from the standard email/xmpp through to the plethora of 
social sites and comments where communications can reach us, there is no 
notion of control, and most importantly no notion of a universal filter 
or universal block - Mrs Green cannot find a way to block all 
communications from her abusive ex partner.

I should end this mail, but certainly thanks for pointing to the video, 
much appreciated.

Best,

Nathan


More information about the foaf-protocols mailing list