Trusting people and machines

Trust is fundamental to human activities. How is it changing?

Would you trust Sophia, a robot that is a citizen of Saudi Arabia?

On a day-to-day basis, people put a lot of trust in others. As I walk down a suburban street, I trust that a driver will follow the curve of the road rather than drive straight into me. The driver trusts the engineers who designed the car that it will not explode, at least not on purpose. Buying an aspirin is premised on trusting the chemists and manufacturers that produced the drug.

            When trust is betrayed, it is a major issue. When, last year in Australia, a few needles were discovered in strawberries and other fruit, it was national news. People normally assume that fruit purchased from a shop has not been tampered with.

            Paedophilia in the churches was covered up for decades. When it was finally exposed, it destroyed a lot of trust in church leadership and the church as an institution.

            Scientific knowledge is based on observation, experiment and theorising, but also relies heavily on trust between scientists, who need to rely on each other to report their findings truthfully. This helps explain the enormous condemnation of scientific fraud, when scientists manipulate or fake their results.

            In certain areas, public trust has plummeted in recent decades: trust in public institutions including government, corporations and the mass media. Opinion polls show large declines. In Australia, trust in financial institutions had been dropping due to scandals, and that was before the royal commission revealed widespread corruption. When people can’t trust their financial advisers, what should they do?

Public trust in Greek institutions has plummeted.

            In order to ensure fairness and good practice, governments set up watchdog bodies such as ombudsmen, environmental protection authorities, anti-corruption commissions and auditor-generals. One of the casualties of the banking royal commission has been the credibility of financial watchdogs such as the Australian Securities & Investment Commission (ASIC). Rather than sniffing out bad practice, they were complacent. Whistleblowers reported problems, but ASIC ignored them. The message is that members of the public cannot rely on watchdog bodies to do their job.

Who can you trust?

Rachel Botsman has written an insightful and engaging book titled Who Can You Trust? She argues that in human history there have been three types of trust.

            First was local trust, based on personal experience in small communities. If someone you know helps, or fails to help, in an hour of need, you can anticipate the same thing in the future. Local trust is still relevant today, in families and friendships. People learn who and when to trust through direct experience.

            Next came institutional trust, in churches, militaries, governments, and professions such as medicine and engineering. People trusted those with greater authority to do the right thing. In the 1950s, high percentages of people in countries such as the US said they had a great deal of trust in their political leaders. However, institutional trust has taken a battering in recent decades.

“So why is trust in so many elite institutions collapsing at the same time? There are three key, somewhat overlapping, reasons: inequality of accountability (certain people are being punished for wrongdoing while others get a leave pass); twilight of elites and authority (the digital age is flattening hierarchies and eroding faith in experts and the rich and powerful); and segregated echo chambers (living in our cultural ghettoes and being deaf to other voices).” (p. 42)

            Botsman writes about the rise of a third type of trust: distributed trust. People trust in systems that involve collective inputs, often anonymous.

Distributed trust

Suppose you want to see a recently released film. If you rely on local trust, you ask your friends what they thought of it. If you rely on institutional trust, you see what the producers say about their own film: read the advertisements. Or you can rely on distributed trust. For example, you can look up the Internet Movie Database (IMDb) and see what different film critics have said about the film, see what audience members have said about the film and see the average rating audiences have given the film.

            If you take into account audience ratings from IMDb, you are trusting in two things. First, you’re assuming that audience members have given honest ratings, and that the film’s promoters aren’t gaming the system. Second, you’re assuming that IMDb’s method of collecting and reporting ratings is honest. After all, IMDb might be getting payoffs from movie producers to alter audience ratings.

            Botsman says distributed trust seems to be reliant on technology but, ultimately, human judgement may be required. Of course, people design systems, so it’s necessary to trust the designers. However, after a while, when systems seem to be working, people forget about the designers and trust the technology.

            One of Botsman’s examples is the self-driving car. Developers have put a lot of effort into figuring out what will make passenger/drivers feel safe in such cars. This sounds challenging. It turns out that the main problem is not building trust, because after being in a self-driving car it seems quite safe. The problem is that drivers become too trusting. Botsman thinks her young children will never learn to drive because self-driving cars will become so common.

            Botsman has a fascinating chapters on the darknet, a part of the Internet frequented by buyers and sellers of illegal goods, among other nefarious activities. Suppose you want to buy some illegal drugs. You scroll through the various sellers and select your choice. How can you be sure you’ll receive the drugs you ordered (rather than adulterated goods) or that the seller won’t just run off with your money and not deliver the drugs? Botsman describes the trust-building mechanisms on the darknet. They include a rating service, rather like Amazon’s, and an escrow process: your payment is held by a third party until you’re satisfied with the goods. These darknet trust-enablers aren’t perfect, but they compare favourably with regular services. It turns out that trust is vital even when illegal goods are being bought and sold, and that reliable systems for building and maintaining trust are possible.

            In Sydney, a high-rise apartment building called the Opal Tower had to be evacuated after cracks were found in the construction. Experts debated when it was safe for residents to return to their units. Some commentators blamed the government’s system for checking compliance to building codes. Could trust in builders be improved by learning from the systems used on the darknet?


Botsman’s special interest is in the blockchain. You might have heard about the electronic currency called bitcoin. Used for purchases online, it can provide anonymity, yet embedded in the code is a complete record of every transaction. Furthermore, this record can be made public and inspected by anyone. It’s as if a bank published online every transaction, with amounts and dates, but without identifying who made them.

            Botsman says bitcoin is a sideshow. The real innovation is the blockchain, the record-keeping code that enables reliable transactions without a middleman, such as a bank, taking a cut. It sounds remarkable, but blockchain-based operations have pitfalls. Botsman describes some disasters. When a new currency system was set up, someone found a glitch in the code and drained $60 million from the currency fund, one third of the total. The programmers and founders of the system were called in to intervene, which they did, preventing the extraction of currency.

            Blockchain seems not quite ready to provide a totally reliable trust system, one not reliant on human intervention. But lots of people are working to achieve this goal, as Botsman revealingly describes.

Trust and political systems

For me, the value of Who Can You Trust? is in highlighting the role of trust in contemporary life, especially as trust in institutions declines drastically. It made me think in a different direction: political alternatives.

Rachel Botsman

            The political philosophy of anarchism is based on the idea of self-management: people collectively make the crucial decisions affecting their lives without systems of hierarchy, namely without governments, corporations or other systems of domination. The usual idea is that there are assemblies, for example of workers who decide how to organise their work and what to produce. Assemblies elect delegates for coordination by higher-level groups.

            This model of self-management relies on two types of trust. The assemblies have to be small enough for dialogue in a meeting and thus rely on local trust. The delegate structure parallels distributed trust, as long as the delegates remain bound by their assemblies and acquire no independent power

            Another model is demarchy, which also dispenses with governments and corporations. In a local community, decision-making is carried out by citizens panels, with maybe 12 to 24 members each, whose members are selected randomly from volunteers. There could be panels for transport, manufacturing, art, education and a host of other topics. In essence, all the issues addressed by governments today are divided according to topic and allocated to randomly selected groups of citizens.

            Because they are randomly selected, panel members have no mandate, so their terms are limited. For coordination, experienced panel members would be elected or randomly chosen for higher-level panels.

            Demarchy relies on local trust, especially on the panels, and on distributed trust, namely trust in the system itself. This distributed trust is similar to the trust we have today in the jury system for criminal justice, in which randomly selected citizens deliberate together and make judgements. People trust a randomly selected person, who has no personal stake in the outcome, more than they are likely to trust a lawyer or a politician.

            Botsman’s analysis of trust and technology raises a fascinating option: what would it mean to combine distributed trust based on technology with the local/distributed trust in political systems like anarchism and demarchy?

Brian Martin