Intro:

I'd like to create a community website, similar to Hackers news's format but for AI Safety enthusiasts. It's more open than LessWrong and Alignment Forum, with a different reputation system and goals.

The idea is to create an accessible space for AI Safety that brings together what’s happening across the community

As on HN, you post a link with a short comment; you don't directly post the article there. Anyone can post, even with a reputation of 0. The following tags will be used: #alignment #interpretability #evals #security #governance #misc

Reputation Points (RP):

RP measures how much your posts are liked. Each vote is always worth 1, with no multipliers. RPs are per tag, not overall, and are divided by comments and posts (12 boxes in total). Each RP is the sum of your five most upvoted posts/comments, striving to prioritize quality over quantity and avoiding profiles boosted by a single lucky viral link. There are no downvotes. The profile displays 12 boxes and total RP, which is the sum of the 12 boxes. Once you reach a certain RP threshold, you become a "member" and can report or endorse posts.

The order on the Home page is by post, and RP has no effect on the ranking: each link increases or decreases only based on the upvotes received and the time elapsed.

Moderation: (report / endorse)

Anyone can post from day one. Links from new accounts enter a public queue and are promoted to the Home page after a certain number of votes or thanks to the endorsement of a "member." Votes from accounts created less than 48 hours ago do not count towards the ranking.

Members can use two opposing gestures, "report" and "endorse": report when content violates a rule, endorse when it respects the rules (endorse ≠ upvote). The user's profile will show how many times they have used them and the percentage of agreement.

After a certain number of clear reports, the content collapses, revealing the reason, exits the Home page, and goes into the "Purgatory" list. It can resurface with new guarantees, alerting the author. Once the decision is made, if the content is valid, it will be moved to the correct list, if it is invalid, it will be removed and sanctions will be applied to the user (Rules: no spam, no off-topic, no duplicates, personal attack, doxxing, illegal, infohazard).

The main risks are low adoption, problems in the moderation system, operating costs exceeding available resources, no enough funds for support development. but even in the event of failure, the code will remain public and reusable.

Technical Part:

Responsive and accesible website following FOSS philosophy, The project is being developed by me, a former Software Engineer, who will directly oversee design, development, and technical management (could be fantastic to have collaborators).

Costs and schedule:

Part-time development: Salary: €1000/month × 12 months and €2,800 will be allocated for hosting, domain, infrastructure, security, tools, and initial operating costs (the €2,800 will increase over time based on adoption, and in that case, funding will be sought through partnerships, never through on-site advertising).