Definition of trust
A simple definition of trust in a social context could be
“the belief that a person or system will behave predictably, even under stress” — source ldapwiki
This definition is further clarified here:
One could also define trust as the expectation of a predictable, repeatable pattern that is subjective to the observer.
Defining trust is very abstract and even may be subjective. However, from a digital perspective, the most simple definition makes it a lot more simple to implement.
The underlying fundamental incentive for trust is security. Trust can be seen as a tool to gain security quickly. Both animals and humans are using “trust” to protect themselves. Trust is a way to skip lengthy security assertions whenever interacting with others. What separates humans from animals is the ability to leverage on others “trusts” in a more complex context.
Rules for gaining trust
A way to gain trust is to repeat the same pattern again and again. To lose trust is to break the pattern unexpectedly, and this usually leads to some pain for the involving parties. Therefore it is generally harder to gain trust than losing trust.
These simple rules enable the development of relatively simple autonomous programs that can still function very efficiently.
The autonomous Taxi
Imagine a fleet of autonomous self-driving electric taxis that individually compete for customers. A customer puts up a ride on auction, and it is executed within a few seconds, and the car with the lowest price wins. When the customer has been delivered to her/his destination, the vehicle will trust the customer based on the behavior on the ride. Customers that make a mess in the car, so it needs cleaning afterward, can find themselves with very low ratings and, therefore, could have a hard time getting another ride later on. The fleet of cars, even that they are autonomous, can choose to share the trust of their customers between other members of the fleet and in this way, avoid difficult customers based on the experience of the fleet.
Because the cars are autonomous and there is no central authority to take care of the vehicles, therefore they need the protection of the network of trust to avoid the risk of getting trashed. Insurance can also be used; however, that would have to be added to the price of the ride based on the customer’s trust rating.
The above scenario is still easy to program, and it does not require any machine learning or AI. (The automated driving software is not considered here).
Global trust
Objectively use of trust makes it fragile to attacks. This is what most e-commerce systems are based on today. When a product gets some reviews on the product website, the entity who wrote the review is usually unknown to you. This makes you dependent on that is no foul play involved, but you have no way of checking.
Conclusion
Trust is the belief that a person or system will behave predictably, even under stress. To gain trust is to repeat the same pattern again and again. To lose trust is to break the pattern unexpectedly. Defining trust can be very subjective, but some use of simple rules would, from a software perspective, be sufficient in most cases.
