Re: [Social-discuss] Control own privacy, posted by _others_

[Story Henry]

On 7 Apr 2010, at 09:19, Matija Šuklje wrote:

> Dne torek 6. aprila 2010 ob 12:22:52 je Story Henry napisal(a):
>> On 6 Apr 2010, at 10:53, Rob Myers wrote:
>>> On Tue, 06 Apr 2010 00:05:32 -0400, Ian Denhardt 
> <address@hidden>
>>> 
>>> wrote:
>>>> Here's the problem I see with this: I'm running a gnu social instance on
>>>> my own server, quite literally a PC sitting under my bed. How do you
>>>> justify saying I can't make your name, as it appears on my website,
>>>> running on my hardware, a link to anywhere I please? Supposing I don't
>>>> have an instance of GNU Social, I just have a website. should I not be
>>>> allowed to manually link to various people, who may or may not want me
>>>> to do so? It's possible it would be impolite of me, but ultimately
>>>> there's a free speech issue there.
> 
> I'm all for free speech, but the thing is that legally as well as IRL you 
> get a clash of privacy vs. free speech. To put it in legalspeek: "One's 
> right extends only as far as another's begins."
> 
> I deliberately put a few quite pervasive methods as examples to provoke a 
> debate, because I feel this is most definitely one of the biggest problems 
> we have to solve.
> 
>>> Possible solutions:
>>> 
>>> 1. Have "anti-tags" that the software respects by default. Or would that
>>> end up being a source of hilarity like Outlook message recall emails to
>>> mailing lists? They would making searching for embarrassments easier than
>>> simply leaving the original tag unchallenged.
>>> 
>>> 2. Allow people to ignore tags from other instances on their instance,
>>> and to not propagate those tags to other instances.
>>> 
>>> 3. Require that tags are confirmed, and simply leave tags unconfirmed on
>>> the other instance if the tagged user declines to confirm them. This
>>> avoids the embarrassment flagging problem of 1.
>> 
>> Yes, this is a bit the way foaf:knows works. You can claim you foaf:know
>> anyone. This does not require them to link back. For a third party an
>> unconfirmed foaf:knows will have less weight (since people can claim they
>> know anyone).
> 
> I'm quite partial to a combination of solutions suggested by Rob Myers, 
> Henry Story, Odin Hørthe Omdal and Ted Smith.

Thanks :-)

> 
> From what I understand (note: IANAC IAAL[1]) it'd be possible to have a 
> trust-bases tag system by using FOAF for user2user trust, and GPG for 
> server2server (and user2user?) trust. The more trust a tag (or user) would 
> have, the more present it would be and vice versa.

note, with foaf+ssl you get a p2p trust based system. There is no need to
distinguish clients and servers in any essential manner.

 http://esw.w3.org/Foaf+ssl/FAQ

I have added a few pictures there now that make it clearer how 
foaf+ssl goes beyond gpg.

> 
> Legally speaking there still can be a problem of e.g Aaron running and 
> maintaining a GNU Social instance where he (or e.g. Ben as a user hosted 
> on it) would write "Chantalle is a whore, here's a bunch of pics" with 
> photgraphic "proof" (whether true or fake). In some cases it's "just" a 
> tort, but in some cases stuff like that can be even a crime. Again: I'm by 
> far not against free speech, but although such stuff is (hopefully still, 
> albeit decreasingly!) borderline cases, we do have to take this potential 
> into account.

That is simple. The person who wrote and published it is legally liable, 
under whatever the rules that end up being worked out for international 
communication are.

> The question is how do we handle it. Of course a possibility 
> would be to state in the ToS[2] that each user is legally responsible for 
> his/her own postings, but we'd have to check how far the host (e.g. Aaron) 
> can waive such responsibility in diverse legal systems.

That is something for lawyers to work out. It's not really an engineering
problem. 

What we do need a bit further down the road is an RDF vocabulary for 
propositional attitudes, so that people can if they publish something
make some claims about what their relation to the proposition is: serious
affirmation, joke, reporting what someone else said, .... That could help
reduce issues such as the Robin Hood airport case

http://www.independent.co.uk/news/uk/home-news/twitter-joke-led-to-terror-act-arrest-and-airport-life-ban-1870913.html


> 
> That being said, it could be possible to solve the problem with some 
> _sane_ ToS and the trust-based tagging system. I agree not everything 
> should be solved by technology (and not everything should be solved by law 
> either). But the FOAF+GPG trust system seems to me like a good example of 
> technology offering a tool for the solution.
> 
> Such a system would also enable trust or reputation of a user and his 
> actions. It could be then up to each GNU Social instance or host (e.g. 
> Aaron) to decide whether to block users who have an abnormally low 
> trust/reputation, just show the users' trust/reputation and let other 
> users' decide whether they trust it or not or completely ignore the 
> reputation system.
> 
> The user reputation system can lead into another problem then. As happened 
> in Second Life, it could happen that mafia-like groups would form who 
> would blackmail people in order not to vote their trust/reputation down. 
> This, I imagine, can be avoided by only allowing trust/reputation to be 
> influenced by people marked as "FOAF:knows".

There is no centralised system here with a centralised algorithm to break. 
There is no one reputation system in this p2p system that is being 
proposed, IMHO. There are just claimed relations. How each node evaluates 
the claims of other nodes will be a subjective thing. Welcome to the 
postmodern database.

  http://blogs.sun.com/bblfish/entry/the_coming_postmodern_era

( a bit tongue in cheek )

Who you trust is completely up to you. If you trust mafia people, then 
you will end up with a lot of mafia type problems. If you trust entirely
what big corporations tell you, you may end up massively obese with only one
muscle in your channel changing finger. 

This is very similar to web pages as they are currently. All we are doing 
is making it more machine readable, so that we can interact more easily. 

Don't get me wrong, truth is still essentially important in fact. I am just
reading Prof. Christoper Peacocke'a recent book "Truly Understood", where he
is arguing very clearly on the importance of truth & reference to understanding 
http://bit.ly/aIFYq0 

We have Reference on the web: the URL.
We have a way to tie users to statements they make: public/private key crypto
We have a reputation system to allow us to tie documents to people, and link 
people: foaf

But we keep it free for each user to decide which community he wants to be 
part of, and whome he trusts. At that level we are following simple economics: 
allow
each person to decide as much as possible, and allow global decisions to
emerge based on micro decisions by many people. 

> 
> I hope what I write makes sense (see [1] again). I really don't want to be 
> the sourpuss, but I think such crucial issues should be tackled beforehand 

Here I disagree. They don't need to be tackled beforehand. If you were right,
then they should have been tackled before the web was created, and timbl would
still be there debating why hyperlinks were important ...


> ...even if they have only limited connection to open source, they have a 
> huge connection with and influence on freedom on the internet!

The internet/web is already here, and functioning. We just need to add the 
pieces
as we find both an audience to use them and as they become practical. Logicall
the Web is extremely well architected. So I don't have any doubt it will be 
easy to
add this when we need it.


Henry


> 
> 
> Cheers,
> Matija
> -.-.-
> [1] I Am Not (much of) A Coder, I Am A Lawyer
> [2] Terms of Service
> -- 
> gsm: +386 41 849 552
> www: http://matija.suklje.name
> xmpp: address@hidden
> 
>