W3C home > Mailing lists > Public > public-xg-webid@w3.org > December 2011

Re: Important Question re. WebID Verifiers & Linked Data

From: Henry Story <henry.story@bblfish.net>
Date: Thu, 22 Dec 2011 11:05:07 +0100
Cc: public-xg-webid@w3.org
Message-Id: <AA15D897-9348-45B0-894E-DF76306B8867@bblfish.net>
To: Kingsley Idehen <kidehen@openlinksw.com>

On 21 Dec 2011, at 19:08, Kingsley Idehen wrote:

> On 12/21/11 12:55 PM, Mo McRoberts wrote:
>> On 21 Dec 2011, at 17:47, Kingsley Idehen wrote:
>> 
>>> I used to think so until Henry expressed questionable suggestions about URI handling that breaks the abstraction re. WebID verifiers.
>> I’m think it was actually Peter initially, but I could be wrong; Henry just revisited the issue, and took a safe (from a security perspective, if broken from a web arch angle) default position.
>> 
>> I’m not sure why that prompted this whole thread. Just saying “redirection (and indirection!) are a fundamental part of web architecture, we just need to settle on how they’re handled from a security perspective” would’ve been a perfectly decent answer to Henry’s question…
>> 
>> M.
>> 
> 
> Here is how I would frame a security problem (something I've done in the past).
> 
> An owl:sameAs relation exists in a graph somewhere along the de-reference trails. A verifier follows the link and finds match. Or said verifier applies inference and makes a union and then gets a match. In either case, one deftly placed relation have tipped the apple cart.
> 
> Solution: implementers of WebID verifiers have to factor in crawl depths and relation semantics. Suggestion could go as far as seeking signed claims for specific relations. BTW -- this doesn't have to be part of the WebID spec, it's just a note for engineers.

I think following owl:sameAs relations is better left to the authorisation part. Requiring it of the Authentication part, makes the entry point into WebID harder, as it adds a bit of a reasoning layer. The spec states that the WebID should point to a WebID  Profile and that this should contain the key.

Now it is true that the spec currently has a pointer from the html to other representations too.

Anyway I have not implemented this part, and I wonder how many people have. This is also not something we have discussed. Perhaps we should open an issue on this one. I can see that there are other things to follow up on here. What about seeAlso links, or alternative relation links, or what perhaps a movedTo relation? But should someone still be using that WebID if they have a new WebID somewhere else? So this opens up a lot of interesting questions.

> 
> The ultimate challenge for WebID is this, you are going to have variation re. product quality. That's fine, a spec can't control actual engineering, it can only provide the specs for the act of engineering.
> 
> The Internet was broken security wise before the WWW came along. WebID has a great shot of fixing this problem, but it really has to understand and honor the age-old practice known as separation of powers.
> 
> The WebID spec shouldn't be about encouraging implementations that are fundamentally technology Camels -- the usual product of attempting innovation by committee. A spec must sit distinct from implementation engineering.

ok.

> 
> -- 
> 
> Regards,
> 
> Kingsley Idehen	
> Founder&  CEO
> OpenLink Software
> Company Web: http://www.openlinksw.com
> Personal Weblog: http://www.openlinksw.com/blog/~kidehen
> Twitter/Identi.ca handle: @kidehen
> Google+ Profile: https://plus.google.com/112399767740508618350/about
> LinkedIn Profile: http://www.linkedin.com/in/kidehen
> 
> 
> 
> 
> 
> 

Social Web Architect
http://bblfish.net/
Received on Thursday, 22 December 2011 10:05:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 22 December 2011 10:05:42 GMT