with the Internet Centralisation of power Even though the Internet started as a truly decentralised network, balance of power has been broken by the dominating services that now support it. A centralised network has the power of information and knowledge in the hands and can potentially dictate what is true and what is false. Unknown provenance of information We all make daily decisions on the basis of information we find on the Internet. But the provenance of this information is hard, slow, and costly to verify and their quality itself is often uneven and unassessed. Someone with no credentials / expertise but with a large community can get high credibility on social networks and ultimately in mainstream media. Information can be corrupted by malicious storage and network, or by censorship and can be shared and propagate to unforeseeable extent.
with the Internet Anonymity & unreliable identities The practice of publishing anonymously or pseudonymously has a long history in different media leading sometime to misinformation. Moreover even though there is no way today to be truly anonymous on the Internet, there is a need to retain at least some amount of anonymity and protect the privacy of the people who need it. Removing anonymity from the Internet is thus not the mean to mitigate information disorder on the Internet. Even with anonymity removed, the issue of misinformation will remain: real people can and will provide false information for different reasons. No fair rewards for good quality contribution Various platforms publicly expose users’ ratings as metadata over the public internet, typically relating to the profile of single users. This model is flawed in two ways: 1- it allows spam to mislead prospective consumers, while past consumers have little incentive in providing their feedback, 2- the revenue that service providers make are not shared with the users that took the time to provide feedback.
with the Internet Bias in Artificial Intelligence software The under-representation of some social groups both in privately owned companies and in governments de facto excludes those groups from contributing to ethical questions and discussions. For example, Amazon’s recruiting engine was shelved because it was shown to unfairly discriminate against potential female hires. Bias in AI systems such as evoked in the aforementioned example create mistrust between humans and machines that learn. AI systems, if not trained correctly can only reveal how human are partial, parochial, and cognitively biased leading us to adopt partial and unequal behaviors. Cyber Threats People and companies are exposed to cyber-crime and fraud in e-commerce. Blockchain transactions are recorded, so that malicious actions can be actually tracked and reversed.
objective • One-way • Transitivity is a common assumption • If Alice trusts Bob and Bob trusts Eve then Bob is expected to trust Eve to some extend, yet not always for interpersonal trust • True for institutional trust • Often trust is built by means of reputation or verifiable credentials
likelihood metric that an entity (sb or sth) possesses a certain property • Based on named or anonymous feedback • Verifiable credentials are tamper-evident claims and metadata that cryptographically prove their issuer • E.g., digital employee identification cards, digital birth certificates, and digital educational certificates • Underlying principle: since trusted authorities issued the credentials, they can be trusted
In fact, trust is distributed among different actors in the system • Economic incentives are provided so that actors follow the rules • The transfer of digital assets is secure • Blockchain maintains distributed ledger of transactions so that double spending is not possible • No security guarantees against fraud or low quality!
• Wallet IDs are not always linked to identities of real people • Moral hazard problem! • The QoS is uncertain: • Market of lemons problem! • No data provenance or credibility • No data identifiers • Law jurisdiction is complex • Information may be hard to find or not scale!
verifiable credentials? • Week identities • Feedback credibility • Biased feedback: Fear of retaliation against negative feedback • How to securely store, retrieve, update reputation in decentralized settings? • How to link reputation to entities and feedback to transactions?
to real people or to trusted entities • But, protect privacy by only using credentials not identity • Build reputation-feedback smart oracles for credible feedback • Link reputation feedback to real blockchain transactions, but unlink it from raters! • Update reputation information based on smart contracts • Break the link between the raters and their feedback • Maintain reputation information for registered entities
built for everything, e.g., data credibility, seller honesty, etc. • Semantics are maintained with the reputation data • Feedback smart oracles can ensure genuine ratings and reviews • No central point of failure • Scalable solution with sustainable economics
ontochain
tamaiyutaro
hirotomotaguchi
emiki
judeeeee
naitosatoshi
recruitengineers
pingcap0315
meihei3
kentaitakura
maroon1st
foursue
databricksjapan
carmenintech
notwaldorf
keathley
reverentgeek
jeffersonlam
philhawksworth
morganepeng
phodgson
jakevdp
shlominoach
jmmastey
moore
