Web of Web Trust Server

Making the web a bit more assessable ⚖️

Idea

The current web is filled with valuable information and can help you in informing yourself. But when informing oneself, you might quickly find out that assessing information and distinguishing between facts and disinformation can be increasingly hard, especially when researching a controversial or political topic.

To help with tackling this problem, we have gone back to the roots of the first search engines and thought about an easy-to-use but informative way of giving the user insight in the reputability of a website. Similar to the first search engines, like yahoo, we assess websites based on the number of times there are linked by other websites and how many times they link to external websites. This method in itself is simple and doesn't really provide a good way of assessing sources, though.

And that's where our special sauce comes into play. By having a seed set for both factual news and misinformative news sites, we can build two networks describing the above-mentioned method. We can use these networks to evaluate a score and present it to the user. We can then supplement the data with metadata about, e.g., topicality and also, since we have the full index, sites that link to the current website. This allows the user to gain deeper insight in the trust between websites and allows us to build a web-of-web-trust that provides explainability and transparency for scores. We also want, (WIP), to allow the user to set their own weights in the composition of the score to allow for a more personalized scoring.

End-Product and Architecture

We plan to realize our idea in the form of a browser extension that allows the user to have immediate feedback when visiting a new site.

See our current progress

Privacy-First Design

Thanks to e1mo and em0lar we also have a privacy first design for information fetching by clients.

Why is a special design even needed?

Since all clients, such as the browser extension, will be fetching a website that the user currently visits, it would be easy for a malicious operator to track all users.

To mitigate this thread, clients must request domains, and ruegen, by the first chars of a BLAKE3 hexdigest of the FQDN. The API will then return all domains that start with the supplied characters in a paginated manner. By using this technique, the server may not know the specific request domain.

Credits

Original Developers from Jugendhackt Berlin 2021:

• Cobalt – GitLab | Website
• e1mo – GitHub
• em0lar – GitHub | Website
• funi0n – GitHub
• NecnivlixAlpaka – GitHub
• Schmensch – GitHub
• smyril42 – GitHub
• Vincent – GitHub
• Wolfaround – GitHub

And our mentor:

• pajowu – GitHub | Website