loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

Thoughts about Trust on a Sunday afternoon.

An awful lot of trust models talk about things like probability, or subjective probability, in terms of how trust works. This has troubled me for a long time. Why? Because trust isn’t probabilistic. Not even a little bit. Not even subjectively. Now I know that Gambetta defines it that way and I know Sztompka defines it in terms of a bet on the future. I know a huge number of trust models papers define it in similar terms (absolutely not all, as Castelfranchi and Falcone’s 2000 HICSS paper makes clear). But the words are all wrong. Now, I wouldn’t dare to suggest that all these eminent researchers are wrong. It’s just that words have power. When we talk about things like probability, even if we finagle the discussion with adding ‘subjective’ to it, it means something. I suggest that we need to think about things differently. In particular, it would serve us well to move on from probabilities altogether. Trust isn’t a probabilistic notion. It’s not like the people in a trusting relationship get up every morning and flip a coin every time they want to make a move. More importantly, when we think about trusting people, we don’t think in terms of them flipping a coin to see if they fancy being trustworthy today or not Plus people are really really bad at probability… So even if I was barking up the wrong tree, it still wouldn’t work. So whilst there’s, objectively, a likelihood that someone might do something, it’s not up to chance. Let’s put it another way: when I’m in a relationship and considering someone being able to do something, whilst I might say that the person is “more likely than not” to do that thing, it's not like they roll a die and do it one day and not the next because the die rolled differently. What’s more, trust behaves entirely differently when we come to the decision to place it. It’s at that point that it changes from an “I wonder” to an “I will”. What do I mean by that? You see, when we decide to ‘trust’ someone to do something, we don’t think about how much we trust them at all. We trust them. That’s it. When we decide to trust the babysitter, we don’t run back from the movie theatre every minute to see if the probability of their behaving properly has changed. When we trust the brain surgeon to do the operation, we don’t keep getting off the operating table to see if they are doing it right. Indeed, when we’ve decided to trust someone and actively done so, the issue of trust is moot. That bit’s done. You can revisit it later if things go right (or wrong), but when you are on the operating table or in the movie theatre it’s just done. This is of vital importance when we think about that Trustworthy AI that is occupying the minds of some very clever people right now. Why? Because if we design systems that think in terms of probabilities like that, they’ll simply get it wrong. They will start doing things like rolling dice, or thinking we do, or worse. It’s not up to chance. It’s up to us. In a future post I’ll tease out what some of this means from the point of view of computational trust, things like zero trust security, and the atomic nature of events. Come along, it’ll be fun!

This needs pointers to the references I make, which I will add in a short while.

Dalkeith, Ontario October 24, 2021

If anyone has been paying attention I started doing this stuff on a Wordpress site in early 2021, but it never really sat well with me. No reasons I can think of. There you are, people are odd. I kind of like the way things look on Universe. I’m aware of the issues, no need to worry at me with them. We’re all good.

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING

loading-indicator

LOADING