[pbhjpbhj]

>We have a pretty good idea of which people are trustworthy (or capable, or dependable, or any other characteristic) in our daily lives //

We really don't. People get surprised all the time that someone had an affair, or cheated, or ripped someone off, or whatever. "But I trusted you" ...

It's actually relatively easy to fool people in to trusting you, as many red team members will probably confirm.

Look at someone like Boris Johnson, people are trusting him to lead the country knowing that he's well known to betray people's trust and that he even had a court case lodged against him based on his very blatant lying to the entire country. You can even watch the video of him being interviewed where the interviewers says (paraphrasing) "but we all know that's a half truth" and BoJo just pushes it and pushes it and refuses to accept that it's anything other than absolute truth.

>If we need to get information from someone we don't know, we form a judgement of their trustworthiness based off of input from people we trust--e.g. giving a reference. //

This is domain authority again - trust some domains manually, let it flow from there. If that domain trusts another domain then they link to it, trust flows to the other domain, and so on. Maintaining such trust for a long time adds to a particular domains trust factor, linking to domains not trusted by others detracts from it.

[pavas]

So how do _you_ make any sort of judgments based off of what people say? What information do you use to judge whether their statements are accurate? Or do you always start with the assumption that everything everyone says is suspect? What sort of information do you use to come to any sort of conclusion, and how do you determine the trustworthiness of that information?

>This is domain authority again - trust some domains manually, let it flow from there. If that domain trusts another domain then they link to it, trust flows to the other domain, and so on. Maintaining such trust for a long time adds to a particular domains trust factor, linking to domains not trusted by others detracts from it.

This can be gamed if you're able to update the trustworthiness of a domain for other people, and that's why a trust metric needs to be mostly personal, and should update dynamically based on your changing trust valuations.

[pbhjpbhj]

Pyrrhonism, you start on the assumption that no-one [else] even exists and go from there ... ;o)

Seriously, I'm not so sure -- I try to trust first and then update that status as more information becomes available; but that's more of a religious position.

I don't think it's necessarily instructive to look at my personal modes here. I guess my main point is that if you're going to say "well humans have cracked trust, we'll just model it on that" then I think you're shooting wide of the mark.