Re: credibility networks (was Re: Is Alice, or her post, credible? (A really rough use case for credibility signals.))

I  very much concur with Connie’s and Sandro’s points, to which I’d like to elaborate or add my own.

The trust networks outlined by Sandro is similar to the “trust chain” idea I outlined in previous messages and calls. I would argue that each path in the graph leading to a leaf (node with no outbound edges) corresponds to a trust chain.
One aspect missing in Sandro’s proposed framework (at leas as I understood it) is the fact that such trust chains (and thus trust network)a are applicable only for a given action and context. This is how understand Connie’s point about “interpretive framework”.
So, for each action+context, the “anchor points” in the chain (ie, nodes with no outbound edges in the graph) will vary. The reasons/justifications for them being anchors can be many, and many studies have already been shown that they are not exclusively based on logical/rational reasoning.

From this point of view, a credibility indicator is any signal I can use to both accept the structure of the narrative (e.g., reasoning and rhetorical style), or, more importantly, where do I place the elements in the narrative in the trust chain. For example, an authenticity signal helps me gauge the identity of the author or of the media included - so any inauthentic element in the narrative can imply in rejection of its claims.

Also, as a consequence, I would argue that there are no universal credibility indicators - I understand this as a conclusion also implied by Connie’s considerations.

I should note that this process happens for each communication (narrative) received. In many cases, when I don’t have enough information to either accept or reject some statements about entities and their relations contained or implied in the claims within the narrative, I will seek additional communications containing statements about these entities and relations. A common criteria to increase my confidence in the veracity of such statements is the number of independent (often very difficult to ascertain!) sources that make similar claims. Here Connie’s caveat applies.

If one accepts the fact that there are no universal indicators, and there may be a very large number of potential indicators depending on the context of their usage, we are faced with the issue already pointed out by several people in this discussion - it is not effective, practical or reasonable to simply surface all possible ones and throw them at the information consumer because we don’t know which ones will be actually used.

It seems to me that one approach to be considered is whether it is possible to somehow surface only “the more important” indicators, but in a customized way -if we have any information about the consumer, or based on some relevant demographics (e.g., location). In addition, additional indicators could be shown on demand from the interested reader - but I suspect only a small fraction of readers would actually request this. I realize this brings in another set of issues, but I don’t see any other way to deal with this problem of multiplicity of potentially relevant indicators.

Cheers
Daniel