Lists Home |
Date Index |
> The example of Edd's "len rating" is a good example of someone
> creating a personal filter for ranking when something is of
> immediate interest vs deferred interest. If Tim Bray writes it,
> he reads it immediately. If Len Bullard writes it, he stashes it.
> That Edd published the rating prominently on xml.com is FOAF at scale.
This seems like where FOAF fits perfectly. If I have a set of people I
deem trustworthy and/or interesting I can use FOAF to indicate that fact
and then a machine can hunt down articles for me with increased accuracy
based on weather the author is a member of my list.
> he had wanted to make it more useful, he could have published an
> FOAF digest which ranked a week of XML-Dev by articles contributed
> and then ranked threads by contributors, links to articles by
> contributors, and so on.
This I understand less well. The idea I get from what you write is that
new FOAF relationships get built between the Edd entity and a number of
articles. My understanding until now is that FOAF indicates relationships
between entities (people or organization).
If Edd had a thing he could write, a kind of "I find interesting" node
(really a relationship between a person and a piece of content), and I
learned to value what he marked this way, I could have my machine feed me
things he marked that way.
If others also created "I find interesting" nodes, I could filter for
articles that had "I find interesting" node that related the article to a
person I trusted. If the "I find interesting" node could have a "because"
attribute and the "trust" relationship had a "with respect to" attribute I
could further filter for articles that people whose opinions I trust with
respect to technology found interesting because they were articles about
the future of XML.
> But FOAF isn't enough. There are articles written by people with whom
> you have no contact and for which you have no personal references.
> The problem is FOAF is the Fox Network (if you are conservative) or
> CNN (if you are liberal).
This sounds to me like a problem with establishing trust chains. Today's
media is all about centralizing trust into a manageable set of trust
relationships and a manageable if sometimes overly generic set of
information channels. People trust ABC news to call the election, Oprah
to recommend books, and Martha Stewart for recipes and decorating hints
(though not necessarily investment advice).
There's no reason authorities couldn't publish FOAF profiles. My
political party could indicate trust for a set of authors and give them a
seal of approval. I could then watch for articles written by (or vetted
by) someone on that list.
There are a couple other ways to draw out this chain. I have a friend
who's opinion I really trust when it comes to other people. If he says
"pay attention to this guy over here, he's really smart" I do. I can
automate this and say "give me articles vetted by anyone my friend trusts
for technical advice". Once trust networks are set up they are subject to
a whole bunch of querying possibilities that just fascinate me.
I can draw out my trust chain in another automated dimension. If I
indicate that I trust the opinion of my political party, my religious
leader, and 16 of my closest friends, I may be able to get a less direct
trust chain back to the author whose articles are consistently vetted and
approved by one or more of those groups.
I'm fairly certain that evidence chains (in the legal sense) could be
modelled using FOAF. Supporting the above scenarios might require you to
allow access to parts of the system that would dilute the rigor somewhat.
I think that if the creation of trust and accuracy meta data became a
natural part of people's browsing habits, that the ability to have
meaningful accuracy gauges on web pages (possibly provided by google or
another aggregator) would quickly become very very good.
> The hard one is evidence chains. Go to court and pull out
> your FOAF files. M. Jackson is doing that. His friends are
> famous and therefore persuasive. This is one reason courts
> in the West are purposely adversarial. In fact, it might be
> interesting to have bots that can take the results of a query
> and break the returns down into adversarial arguments to handle
> the non-monotonic aspects devoid of persons who wrote them.
> But when a chain of factual evidence is established, it will rely on
> verification, traceability, multiple sources, and logical consistency.
> Aggregators are just aggregators. A link is not a virus and
> a tipping point is not reached through a simple numeric threshold
> (that is what is wrong with pure connectionist philosophies based
> on simplified notions of power laws).
> It is the filters you want to be cognizant of
> which can include the aggregating query (see example:
> "anyword" + "definition") or the FOAF rankings.
> The strength of the vector is important but it has an
> application context. FOAF is one that may or may not
> be reliable.
> For a system such as the one described in the original
> article, traceability is far more important because opinions
> don't matter. That is why property/evidence systems have
> scrupulous chain of custody requirements.
> From: Nathan Young [mailto:firstname.lastname@example.org]
> When filtering articles you want to be able to say "show me only articles
> that have been tagged as accurate by people I trust"
> Establishing the reliability of the source is the first part of the
> The xml-dev list is sponsored by XML.org <http://www.xml.org>, an
> initiative of OASIS <http://www.oasis-open.org>
> The list archives are at http://lists.xml.org/archives/xml-dev/
> To subscribe or unsubscribe from this list use the subscription
> manager: <http://www.oasis-open.org/mlmanage/index.php>