There’s a growing backlash towards AI-generative content. AI slop is taking over social media, it’s being used for political propaganda, and it’s even starting to spill over into the real world.
Consumers prefer human-made content, and marketers know this, but it can be hard for folks to quickly identify when a piece of art was made by a real person or by AI.
This is why we need the Human-Made Project, an easy way to quickly certify work that was created without the use of AI.
Golem is a demonstration of how to distribute content over ActivityPub securely over peer to peer networks.
The problems this demo is trying to address are:
- Nodes go down, and their content tends to go with them. How can we have content that survives? Content which is distributable over a peer to peer network seems like it would help.
- Except if an entire network is helping hold onto and distribute content, how do we keep private content private?
- How to do this in a way that is compatible with the ActivityPub specification?
By encrypting the file and splitting it into chunks distributed through the network and only sharing the decryption key with the intended recipient, and by using a URI scheme that captures the appropriate information, we can accomplish all the above. Golem uses the magenc extension of the magnet URI scheme to accomplish the above.
A huge blocklist of sites that contain AI generated content, for the purposes of cleaning image search engines (Google Search, DuckDuckGo, and Bing) with uBlock Origin or uBlacklist. There is also a Pi-Hole compatible list in the repo.
list.txt can probably be processed and used to build a blocking database for search bots.
Much social software doesn't let you manage replies, whether coarsely or in a fine-grained manner. Sometimes you just want to spout off!
Sometimes you don't actually want people to reply.
Chupacabra enables users to archive and discuss web content free of surveilance and commercial influence. It can be used for personal research, micro-blogging, or discussing dank memes. Chupa posts are standalone archives of web content (a single HTML file with images embedded and scripts removed) and a corresponding Matrix message pointing to the mxc:// URI where the archive can be fetched. Posts can be discussed in real-time in the channel that they were shared. Behind the scenes, all post discussion is composed of replies to the post's Matrix message.
The Public Domain Manifesto.
The Cutting Room Floor is a wiki of easter eggs and unused but not removed stuff hidden in video games. All sorts of weird and wonderful trivia about video games can be found here.
The indieweb is a movement in which people own their data, and run their own applications rather than other people's walled gardens. content is there and isn't censored arbitrarily. federation with some or all services is done with interoperable-by-design protocols and data formats. User interaction (replies, likes, stuff like that) are also federated back to your site(s).
The homepage of the Open Graph protocol, a system of tags for turning web pages into smart objects in a social graph. Describes what kind of content it is (which implies how to treat it) and allows functionality to be imported from social media sites. Supplies much greater context for web pages. Includes multimedia content.
DocNow responds to the public's use of social media for chronicling historically significant events as well as demand from scholars, students, and archivists, among others, seeking a user-friendly means of collecting and preserving this type of digital content.
python module for extracting text from different documents. Can also be used as a CLI utility. Can work with text-based formats like CSV, JSON, and HTML. Can work with binary formats like MS Word, MP3, and PDF. The list is fairly extensive.
The main thrust of the article is how to whip together a quick and dirty HTTP server for static pages using Python and Cherrypy. I saved this article because the worst-ranked response also tells you how to hardcode (read, make configurable at runtime) the port and IP address a CherryPy instance listens on.
cherrypy.server.socket_port = 80
cherrypy.server.socket_host = '0.0.0.0'