Technical Side of SpaceHey

Disclaimer for Moderation

Recently, someone used automated accounts to mass-upvote blogs on this platform. I want to be explicitly clear: this was not me, and I have no affiliation with whoever is responsible. I mention this only because what I'm about to describe involves similar technical knowledge, and I don't want the two conflated.

With that being said, here is my proposal.

I'm a software developer, and I've been using SpaceHey for about two years. I know this platform well, not just as a user, but technically. I understand the frontend and the underlying structure at a level most users don't.

You've probably seen the blogs lately. The petitions. The callout posts. People are frustrated. Most of what's being written comes from a user perspective though, so I want to give you the technical one.

SpaceHey has over 2 million registered accounts. Realistically maybe 10,000 are actually active. It has only a couple of moderators. For a platform this size you'd want at minimum a dozen, each handling specific areas like spam, harassment, illegal content, and forums.

I want to be clear that I respect the moderation. Most of the stuff you don't want on here disappears within 15 minutes. Most.

The anti-bot measures on this platform provide no meaningful protection. I won't go into specifics, but this isn't speculation. The proof is the automated accounts already operating on this platform right now, by people who are not me, doing things I don't condone. Anyone with moderate technical knowledge can automate account creation on SpaceHey. The infrastructure to cause harm already exists and is already being used.

About six months ago I reached out to An directly, offering to build an automated moderation system. I got no response. I built it anyway, but did not deploy it for obvious reasons.

The system crawls SpaceHey continuously, runs content through a locally hosted open source AI model, and generates a risk rating per post. I want to be very clear about what gets stored: after roughly 15 seconds, every piece of user data is deleted. The only thing that remains is a post ID mapped to a rating, in a Redis database. Nothing else.

I chose not to deploy this as an automated moderation tool. Taking unilateral action on a platform I don't own, with no transparency or appeal process for affected users, would make me part of the problem.

Instead, I'm releasing a TamperMonkey script, when greenlit, that fetches these ratings through a public API and hides flagged posts for users who choose to install it. The API is public, so if you want to build your own scripts on top of it, go ahead.

An, moderation, if you see this, I'm not your enemy. Reach out via SpaceHey or email me at me@abgespeichert.com if you have any issues, considerations, or thoughts. I will take them seriously.

Shoutout to David, Dana Scully and of course An.

EDIT:

This is how it would mainly look like. It will be a TamperMonkey script, which then interfaces with a public api, so other people can make their own scripts. I will also make a direct js integration.

https://ibb.co/qLSWjv9k

EDIT²

The first version is now fully working. ( without UI ), it fetches the blog violations from a public api, and then displays that. The processing is done by a fully self-hosted llm. 

You can apply as beta testers in the following weeks.

https://ibb.co/1tKvWwC8