opt-out, acknowledgment and FOAF piracy

[Liz Turner <elizaturn at y...>]

--- In rdfweb-dev at yahoogroups.com, Julian Bond <julian_bond at v...> wrote:

> Speaking for myself.
> I see absolutely no problem with my FOAF file saying that I know XXX 
> with mbox_sha1 of YYY. I'm stating a fact here about myself not about 
> them. Even the mbox_sha1 is only saying that *I* can contact somebody 
> called XXX via the mbox who's SHA1 is YYY. It's not actually saying that 
> YYY belongs to them or anything much about XXX at all.

If you assert that data about you is owned soley by you, then surely you must 
agree that data about others is not your property - it's the property of the 
person it describes. 

For instance, I have an email address belonging to one Bruce Sterling. I 
believe it's his, because he sends me email from it. There's nothing to stop me 
from posting in my FOAF data that he's a good friend of mine. People 
reviewing this data might be tempted to take this assertion at face value, and 
imagine that Bruce and I are pals, when in fact we are not - I'm just a fan. If 
Bruce has FOAF data in tha same universe, unless he's looking in the right 
place, he might not be aware that I am fibbing about my relationship with him, 
and the fallacy goes unchallenged.

When we're building tools to present this information, I think it's extremely 
important that people can have some level of trust in the information they're 
looking at, and the claims others make about them. If this is not the case, then 
FOAF is not going to become the useful tool it deserves to be. 

While you may be trusted to act truthfully and ethically in respect of other 
people's data, you must accept that that won't always be the case, and that 
some people are bound to abuse it. 

> The problem arises when we present this as a two way link. A lot of our 
> tools present this information as if it is two way and that YYY says 
> they know me as well. But they never did say that.
> 
> I'd much prefer that the tools displayed the links as one way than that 
> we got into trying to construct a system that guaranteed that the links 
> were two way.
> 
> If we have to go down the route of qualifying links in both directions, 
> I'd rather we reversed the logic and allowed people to say "I don't know 
> this person". The tools would then show the conflict.

This is obviously the simplest solution. It's clearly not desirable at this stage to 
"ignore" all but two-way links, and I'm not suggesting that we should. But I do 
think it's very important that FOAF users are aware of the problems that _not_ 
acknowledging links are likely to cause.

I still haven't thought of a way of graphically representing the subtlties of 
human relationships as a line on a map, and I understand Jim is still 
struggling with the directional aspect in FOAFnaut, so it's my opinion that in 
the meantime we should do everything we can to encourage people to 
confirm or deny their asserted FOAF connections.

> Which all comes back to a belief that my foaf file is owned by me, 
> created by me and describes me. And any information it provides about 
> other people is hearsay and not fact.

In which case, there's hardly any point in using FOAF at all :)

cheers
e