Principles

Credibility is collective

Society runs on trust. We all need to trust that the random people we interact with will cooperate. ... If too many people steal or too many people don’t pay their taxes, society no longer works.

-- Bruce Schneier, Trust and Society

No one authority can define credibility. It's a balance of personal judgement and group norms. Signs and signals will evolve with our culture.

Principle: any credibility score needs to adapt to collective input.

Algorithms are convenient

However, standard credibility assessments provide market efficiencies. Few have time to perform deep due diligence on every transaction. Traditional market (tradfi) institutions like the SEC and credit report bureaus ensure a minimum threshold via licensing, regulation, and monitoring. Meanwhile ratings and reviews allow web2 consumers to "check for themselves" but still rely on centralized systems.

Principle: a credibility score needs to be legible and concise.

Credibility is power

High credibility brings financial rewards. Defining credibility in a market grants influence.

How that power is used depends on incentives. Market makers (Uber, eBay) benefit from trust in sellers and buyers. Some, like Amazon, extract fees from sellers to hijack credibility via paid reviews and ad-based placement. Others, like Yelp and Glassdoor, repeatedly face accusations of extorting businesses by suggesting paid plans may help dispute low ratings.

Principle: avoid centralized incentives to tamper with credibility for profit.

Algorithms change behavior

People will shape their behavior according to what a credibility algorithm rewards. People's livelihood depends on how a centralized algorithm ranks their credibility. When these algorithms are centralized and owned by organizations, obtuse changes can alter careers and fortunes. For example, YouTubers are in a constant cat-and-mouse game reverse engineering the monetization algorithms.

Principle: one must have the ability to influence an algorithm that impacts their livelihood.

Algorithms will be abused

"the optimal amount of fraud is greater than zero."

-- Patrick McKenzie, Bits about Money

Any summary will omit details, so credibility scores can't fully represent the truth. This wiggle room allows abusers to optimize to the point where it defeats the original purpose. Centralized management historically fails; there is no way to perfect the algorithm or police all fraud. Instead, rely on the people most impacted to define the trade-offs between usefulness and abuse.

Principle: collective governance will ensure a credibility score reflects effective trade-offs between usefulness and abuse.