Between 2014 and 2018 there was a rush of books and articles about "trust", some of it prompted by the creeping invasiveness of consumer surveillance, some by social media, some by the surprising results of the Brexit referendum in the UK and the US elections in 2016. We had the economics of trust, the neuroscience of trust, trust for leadership, trust for brand relations, trust as civic infrastructure. There was hand-wringing about how to foster trust, rebuild it, understand it. Most of this was focused on how to get people to trust some thing. Can we get people to trust our brand or this result? Can I get co-workers to trust my plan? Can we get people to trust this government? Those are all excellent and continually relevant questions, I understand why people asked them then and I see why audiences were interested in the answers. But eight years ago feels like an eternity. Questions about trusting a brand feel quaintly naive in the face of political chaos, economic chaos, existential threats from AI, and a rapidly escalating climate crisis.
But I can't stop thinking about the less focused questions of trust because I hear them circling in every podcast about AI, about the future of democracy, about global capitalism. What does it mean for AI to be aligned and can we trust that alignment? What political party should or shouldn't we trust? Which public figures deserve trust and does not? These ask questions about trust without understanding one of the most fundamental features of it: most of the times that we trust, we don't ask whether we should. When we ask whether we should, we tend not to. Imagine that you buy a ticket on an airlines website and the check-out screen says "trust us, your flight won't crash". I, creature of the 21st century, simply assume that flights will not crash even though for a long time they did not infrequently. Pointing out that I assume that feels in and of itself deeply untrustworthy. Imagine living in a place where drivers simply continue through red traffic lights without a care in the world (or perhaps you already live in one). I, creature of the United States, simply assume that I don't need to pay any attention to the cross traffic when I come to a green light. Telling me that traffic lights are not inviolable social contracts, that I in fact do need to slow down or even stop at intersections, feels deeply unsettling. Making me reconsider an action or fact that I did not previously feel the need to consider isn't an opportunity to decide to trust or not, it's the end of trust and the beginning of deliberation. I'm not really trusting, I am deliberating and electing to cooperate in this particular situation. The next time around I'll need to repeat the whole process.
Most accounts of 'trust' confuse trusting with cooperating. This is where the lens of game theory becomes useful because we should think of trust beginning more or less where games end. Game theory gets a reputation for being a dull defense of selfishness because the typical hand-wavy account of game theory explains prisoners' dilemma and then concludes with "betray always wins". That's not even true for the prisoner's dilemma, nor does it reflect the larger lessons that can be drawn from the field. As David Sloan Wilson says: "Selfishness beats altruism within groups. Altruistic groups beat selfish groups. Everything else is commentary." How one fosters altruism for long-term collective gain against the obvious short-term individual gain of selfishness is what political theorists have debated for millennia. Any peak at the news or the current political discourse shows that there is no answer to it. However, the way one behaves once they are in an altruistic group does have a simple answer: for the most part, they blindly and automatically trust.
What people want is cooperation without deliberation because it allows me to do so much more than I could if I had to deliberate. I want to assume that my classmate won't steal my lunch, my husband/wife won't steal my credit card, the airplane won't crash, red light will be obeyed, and so on. A deeply interconnected and technologically automated world requires that we trust because we cannot comprehend the complexity that surrounds us. If we had to deliberate at every decision we would be paralyzed by the number of systems have to interact with on a daily basis. We trust so that we don't have to deliberate because deliberation is slow and difficult. Our world requires us to trust so many entities and institutions that we can rarely get through a day without an unconsidered or even unconscious trust. A modern life quite simply doesn't afford us time to deliberate very often. Trust diminishes the need for information.
The question of trusting also is typically framed as an act between two people. I decide to trust you because of things that I know about you. But many times trust is more a matter of trusting some thing else rather than trusting another person. I don't trust "this plane" or "that pilot". I trust a pilot, a pilot training systems, the airline, the airplane manufacturer, the Federal Aviation Association, the criminal justice system, and an exhausting list of assumptions that I make. There is a constellation of entities and institutions that I assume will all collaborate with one another in a way that I can trust. Even when we trust an individual we're often trusting an instrument that can be used to guarantee their trustworthiness. I have a signed contract, I have faith in a legal system, in a parent company, I know that our mutual friends may shun them. These kinds of trust are not founded in weighing the ethics of another, they're founded in believing that some institution of some kind, be it in a court system or mob justice, will ensure the behavior of that other person.
Those of us who live in the developed post-industrial world live in what Seth D. Kaplan calls "thin societies". These emphasize the autonomy of an individual and direct relationships between those individuals and institutions that respect rights and specific wishes and are based in universal moralities and minimal interpersonal obligations. These contrast with "thick societies", that value deep and unchanging interpersonal relationships bound by tradition, social order, and hierarchy. We might imagine that "trust" in a thin society makes us like those thick societies, but that isn't the case. Thin societies may have mafias, religious sects, utopian communes, that are thick but those are the exception rather than the rule. We don't suddenly enter that kind of society when we cover our friends' dinner or ask someone to keep an eye on our backpack while we run to the bathroom.
I'm interested in trust because, like a lot of people, I work in technology and because I'm interested in building technologies to support humans making decisions. Deciding means deliberating on one feature of a decision point or interaction. Having to deliberate on all features is wildly inefficient and exhausting. Economists are right that we seek efficiency but we do so except in the things that we value. In matters we value, in the games that we want to play, we don't mind slowness or waste. We value it as craft, care, intelligence, and thoughtfulness. Thin societies allow us to delegate almost everything that we don't value enough for deliberation to another institution. There is a caveat to this though: we can only do this if we truly trust the institution. We can only fully delegate by saying, in essence, I don't need to deliberate on whether I will cooperate and whether this institution will cooperate.
Accounts of trust tend to paint a picture of rational actors eyeing another across a poker table or a trader in a market biting a gold coin. These are rich scenarios in which to study human behavior but they're infrequent. What is far more common, what constitutes how we live together, is non-deliberative trust that we don't consider. Confusing choosing to cooperate with trust, be it in a hierarchy, deity, or institution, confuses what we actually do when we trust.