Re: "tag:" Identification Idea [was: Re: Proposal: 'tag' URIs]

Sean,

I'm confused by your proposal in that you seem to advocate an email 
exchange per _tag_. That's highly undesirable and against the 'negligible 
minting cost' requirement in our proposal, when it is at most needed per 
_authority_ -- and I'm not sure it's always needed for that. A minting 
entity doesn't need to register its specific identifiers anywhere -- except 
locally, so that it keeps track of which tags it has minted already. As I 
mentioned before, some freeware for doing this as a local web site would be 
really nice.

(While I think of it: we might find it useful for minting entities to 
publish the syntax of their specific identifiers somewhere well known, for 
software that wants to check their tags' syntax in detail.)

It seems to me that I should be able to mint, e.g., 
tag:timothy@hpl.hp.com;1:blah today, without consultation with anyone else. 
That, of course, is the point. The problem is when a dispute arises: I mint 
tag:timothy@hpl.hp.com;1:blah (which, actually, I have the right to do) but 
someone else says they owned timothy@hpl.hp.com at that date. Whom can I 
appeal to?

It seems to me that something like your idea could be made to work for 
settling that dispute (is this what you intended?):

I sent an email containing tag:timothy@hpl.hp.com;1 to www-uri-tag@w3.org 
on 2001/1/. That registry sent me a signed statement saying 'the possessor 
of value v has the right to mint tags of authority timothy@hpl.hp.com;1', 
where v is an effectively unguessable one-off value that was returned 
encrypted in the same email, so that only I can read it. Now I have a 
capability (in the security sense) for that authority: if I can satisfy 
www-uri-tag@w3.org that I possess v (which I can easily do), it can affirm 
my right to the authority.

I can fill in the security details but I'm sure you get the main thought.

Or did I misunderstand what you were saying?

Tim.

At 04:00 PM 4/28/01 +0100, Sean B. Palmer wrote:
>Many people don't have their own domain name, but do have an email
>address, so I envisage that many people will be minting "email" type
>tags rather than domain ones. Problem is, it is very difficult indeed
>to make sure that someone was using a particular email address, unless
>they send a letter to a list on a certain date... The idea: the W3C
>(or some trusted non-partial entity) could set up a mailing list
>whereby one sends an email with a subject that is the "body" of the
>email tag to be minted, and the list sends back a confirmation email,
>a bit like the Yahoogroups subscription process. Then, as long as the
>tag hasn't already been taken on that date, the W3C would archive the
>tag.
>
>For example, if I sent an email to www-uri-tag@w3.org today, with the
>subject "sbp", and the body "This identifies the owner of this email
>address on this particular date, Sean B. Palmer", then it would be
>sent to the list, which would check that I haven't already requested
>"sbp" today. If I haven't, it would send a message to
>sean@mysterylights.com asking for confirmation. I would reply, and it
>would archive the tag, with a pointer in the archives to
>tag://sean@mysterylights.com;1-4-28/sbp which when referenced would
>bring up the message body.
>
>The only problem with this is that it's difficult to do namespaces, so
>there could be another list (e.g. www-uri-tag-ns@w3.org) that when a
>message is sent to, acts as a namespace. For example, a request for
>"barnyard/" would mean that I could use
>tag://sean@mysterylights.com;1-4-28/barnyard/* as a namespace for
>barnyard animals, rather than having to register barnyard/Horse
>barnyard/Sheep barnyard/Cow and so on separately.
>
>It may also be the case that the W3C would want a (possibly member
>only) list for delegation of tags under tag://w3.org;*/* because I
>expect that certain SW folk will be using them, and there'll be no way
>of keeping track unless some system is se up, even if the minting of
>tags was restricted to Team members.
>
>The more I think about "tags", the more intrigued I become... Good
>idea guys.
>
>--
>Kindest Regards,
>Sean B. Palmer
>@prefix : <http://webns.net/roughterms/> .
>:Sean :hasHomepage <http://purl.org/net/sbp/> .


Tim Kindberg

internet & mobile systems lab  hewlett-packard laboratories
1501 page mill road, ms 1u-17
palo alto
ca 94304-1126
usa

www.champignon.net/TimKindberg/
timothy@hpl.hp.com
voice +1 650 857 5609
fax +1 650 857 2358

Received on Saturday, 28 April 2001 14:48:37 UTC