A User-Value-Oriented Discussion Site - bleep

not written using "AI" or LLMs.

Rut-roh raggy, they mentioned "Value"!

I do not like the way some corporations rely on unpaid volunteers to provide value for their customers. Some people might argue that the power moderators have and the enjoyment they get from managing a community, for instance, compensates them enough. I reject that on the basis that if I wanted to labour for fun, I would go play a computer game or walk my dog, and even some computer games the labour can become onerous to the point of it feeling like an unpaid job.

Therefore I'd like to see a site where people get paid for the effort they put in. The value they provide to a community.

I also think that users should own what they make. Owning what you make unless you explicitly sell it to someone should be obvious, but corporate rent seeking has promoted the idea that nothing is ever really yours.

The ATProto ecosystem of software and services is built on users owning their data and being able to move it to a new host whenever they want. I personally have done this, by moving my Bluesky data to a new host. I still use Bluesky and this portability has inspired me to wonder if we could do the same thing with user groups and comments on a discussion site.

Therefore, what if the discussion groups themselves were ATProto accounts that could be moved between hosts? What if, just like Bluesky, the user data could be displayed and interacted with from different frontends?

Taking these 2 concepts together: Users get paid for what they contribute, and users own what they contribute. Can we make a discussion site that does this?

I think it's certainly possible. Whether or not it's something that people will adopt is something I have little control over, but I think fun could be had in the making of such a thing.

Introducing!

Yeah I don't actually have any site ready yet. 😑

This is my statement of intent: that I want to try to write a discussion site that puts users and creators first.

I don't know if I will want help. Certainly I'm not a prolific coder. If you want to take the concepts mentioned on this page and make a great webapp, please go ahead, in my opinion there's nothing on this page that could be considered original.

Why Discussions?

Bluesky et al already do microblogging. I really don't like the enshittification of things and forums that once down, disappear forever. I want something that can stand the test of time, and if the frontends go down, other people can display the data that exists or move it to other formats.

Features I'd Like

Rich text markup, think more Typst than Markdown, but plain text is just fine too. If pressed I will admit this is purely because I want footnotes and references. 😍
Equitable and welcoming to diverse and marginalized groups of people with a clear code of conduct and safety policy. i.e. don't make the mistakes many feel that other social media has made.
Private accounts and discussion groups.
Astroturfing detection. If an account is sponsored or corporate, it should be clearly marked as such.
Sealioning, trolling, and other anti-social behaviour pattern recognition and prevention.
Subscription based, so no ads.
No user tracking beyond that needed to function.
Transparency of funds and their allocation.
Rich Moderator tools with an API for extending them.
As little client-side Javascript as possible for displaying discussions. We run far too much crap on far too many pages.
No damned 3rd party tracking bullshit.
Use ATProto services for videos and pictures, including pckt.blog!
The better features of Bluesky, like blocking.
Multiple front-ends, much the same as Bluesky/Eurosky/Blacksky and all the others.
Automatic translation.
Clear disclosure of when and how Machine Learning, "AI", or LLMs, are used.
...and many more that I am sure I will think of once I hit "Publish".

Regarding AI

I wanted to use a local LLM to develop this idea. There's a lot of discourse on Bluesky surrounding LLM usage and most of it is very negative regarding LLM use. Which is completely understandable considering how unethically the main producers have acted.

My personal position thus far is that I won't use a cloud-based model because of the associated pollution, fascism, water and electricity usage, and handing over more money and power to billionaires, but I do think local LLMs have the potential to be useful. If I could use one that was guaranteed not to have stolen from content creators, I would do that, although I (currently) don't trust any corporation or state organization to honestly represent the source of their models' training data.

I would also want governments to clearly legislate and enforce that AI corporations must identify and prevent AI psychosis.

Therefore, I can't with clear conscience use even a local LLM. Which really sucks because as someone with AuDHD, pushing through personal projects is very, very difficult. An LLM can provide the structure and novelty that my brain seems to crave for non-work design and coding.

Thank you for reading

-Anthony

PS. As someone with AuDHD I am very scattered. I welcome constructive criticism or even someone just reading this page!