Trust networks

A trust network is a filtering arrangement rather than an affective one, and what is being relied on is knowledge and judgment rather than goodwill. The problem it addresses is the volume of information no individual can assess, which is the singularity condition, and the alternative arrangements both fail in Marsh's account. Certification by an external authority produces an oligarchy that cannot be removed, and popularity produces influence held by social media celebrities using strategies unrelated to knowledge, which is why she says there is an urgent need to separate expertise from popularity. A trust network distributes the assessment instead, with each person relying on the judgment of those whose judgment they have reason to rely on. It fits the framework's other mechanisms directly. Concentric circles promote peers on reputation rather than credential, and promoting ideas rather than people allows an idea to be audited without the distraction of attached personalities. Her own work on the subject was practical rather than theoretical, since much of her thinking about trust networks and reputation came from developing a prototype for Getgee, a project initiated as the Global Square, and from watching people struggle to reach consensus and needing tools that would filter signal from noise. The expanded network has a second function she records. It provides an avenue of appeal where a person feels their local society is shunning them unjustly or their basic human rights are being violated.

Web of trust models in cryptography, notably PGP, implement decentralized trust through chains of personal endorsement rather than certificate authorities. Reputation systems on platforms and in open source projects perform similar filtering, with well documented problems of gaming and of early advantage compounding. Sociology of trust after Niklas Luhmann treats trust as a mechanism for reducing complexity, which is precisely the function Marsh gives it. Her contribution is to tie it to the separation of expertise from popularity, which makes the network a governance component rather than a convenience, since the same structure that filters information also determines who is amplified.