Re: [saag] Liking Linkability

On 15 Oct 2012, at 03:46, Karl Dubost <karld@opera.com> wrote:

> Henry,
> (massive cross posting [bad], reducing to public-privacy)

(this looks like a nice example of a binary judgement that does not take 
context into account.  ;-) Having said that I am ok, with reducing
the conversation to the beautifully named public-privacy list )

> 
> Le 9 oct. 2012 à 09:19, Henry Story a écrit :
>> yes, but don't make linkability an a priori bad thing, since it is the most important  building block for creating distributed co-operative structures, and so to privacy.
> 
> I'm not sure people are making it a bad thing, quite the opposite. People love to be connected with others.
> 
> The issue is to make all these issues a set of binary constraints. I'm linked. I'm not linked. I'm not known. I'm known. etc. It's why I refer to a continuum of opacities from 0 to 1 when I talk about privacy.

we agree. With WebID+Linked Data+HTTP Access control, you can distinguish those relations you wish to be seen from those you wish to protect, in order to build such a continuum.

> 
> What you seem to be talking about is 
> 
> The sealed shared communications of a group of people where people can be >= 2 at *an instant t*.
> 
> But this would be quite a too simple characterization of privacy. We live in time and… WE FORGET. This feature is also very important.

Indeed. Another feature is that we may discover that we wish to remove trust from a group of people. This also works better in a linked web, since I can remove someone from the group of people that have access to a resource (be it read, write or even execute). Of course I cannot
remove the copy those people have of the data, just like I cannot demand you to forget what I have
told you. We can on the other hand add metadata to our resources that indicate what we distribution rules we wish the data to be published under - work Oshani Seneviratne is developing at MIT.

> The bi-directional linkability is about mutual exchange of tokens that we are talking again and again to the same person.

An important point of clarification: in the Linked Data web we are not basing our relationships on bidirectionally. I can express that I foaf:know someone, but that person does not have to reciprocate. Each person/agent is in charge of the statements they make (be they public or private)

> Why it is important? Because through time, we LEARN to build trust with this other person. Little by little. Basically we create a box for privacy through time. It doesn't exist by itself. The more we have tested this mutual trust, the more we are eager to exchange in the context of this box. Sometimes there are small catastrophes. Breach of trust. But the good thing is that we know voluntary or not and depending on the people how to forget allowing more trials for trust.

I agree there too. 

> 
> 
> 
> 
> -- 
> Karl Dubost - http://dev.opera.com/
> Developer Relations, Opera Software
> 

Social Web Architect
http://bblfish.net/