I’ve been doing a lot of work around this area and the issue tends to pop up around context. There are instances a haiku model will catch something far more accurately than gpt-oss-safeguard. Obviously you pay in costs, realistically, you need a full system here for a proper implementation so some flows to tiny LLM some flows exclusively ML, some flows to a higher intelligence.

People on the fediverse are generally pretty anti ai but it’s basically impossible to scale a platform without AI moderation. I would fully welcome any instance trying an implementation of [Osprey](https://discord.com/blog/osprey-open-sourcing-our-rule-engine with LLMs.

This actually ended up not being an external DDOS. They shipped some code that ended up hammering their servers.

What’s crazy is I can’t even pinpoint what they even mean by the wokeness? It’s really not overbearing, nor anymore than SNW or other TV shows. Maybe I’m just in a West Coast echo chamber

In my experience with windows there’s just a slight lagginess everywhere. I’ve had full gaming PCs still feel laggy just in Vscode. It’s not bad but it’s a small pain point that I don’t want to experience for 40 hours a week.

Most of this work is done via emulators if you have a hardware attestation process built in, it will stop most of it. Obviously you still have a problem with phone farms but those are much more expensive than emulators and there’s a physical capacity to them.

I agree the Fediverse is not in a good spot for something like this.

I actually think this is where the identity system of atproto will be more impactful here as it allows a better verification system. I’ve been thinking lately you should be able to use hardware attestation + biometric attestation on apps to filter these emulated users out.

The Hot ordering is itself an algorithm.

Me when Claude code writes bullshit I don’t need

I used to think about this via mesh networks as simply routers, but now with nostr, IPFS, atProto and that new BT messaging stuff Jack Dorsey is on. Technically you could utilize your phone as an access point to the mesh network as you move around the city and load all the comms in the background. The latency would be high, but it could work. Also with 5g tech nowadays long range mesh networks are much more feasible albeit probably expensive for a hobbyist.

There are other Appviews on the atProto that have chosen not to implement checks that have full access to Bluesky posting/data.

Im thinking of the fork the devs of the programming.dev instance were trying to build, it was PHP. Can’t remember the name, then this and Pixelfed. I don’t know if favoritism is the right word but I do feel like 3 projects is a lot for the current ecosystem lol.

Certainly an interesting decision.

What’s up with ActivityPub projects being beholden to PHP. Not even pitting languages against each other, just interesting that most new projects nowadays will choose something like Typescript + React but fediverse tends to favor PHP.

This doesn’t include which models or prompts given in the article. Really need to include that if they have anything worth saying, otherwise it’s just a marketing article for their platform.

This really outlines why I swapped off coffee and started doing two cups of tea instead. Oolong is great

Or even better B complex!

Negativity of the AI crowd is so crazy to me while I’m having Claude Code Opus implement entire UIs on flutter and backend Go interfaces w/ unit and integration tests on a single go.

I have nothing to prove to you if you wish to keep doing everything by hand that’s fine.

But there are plenty of engineers l3 and beyond including myself using this to lighten their workload daily and acting like that isn’t the case is just arguing in bad faith or you don’t work in the industry.

Four lines? Let’s have realistic discussions, you’re just intentionally arguing in bad faith or extremely bad at prompting AI.

It’s possible the library you’re using doesn’t have enough training data attached to it.

I use AI with python for hundreds line data engineering tasks and it nails it frequently.

Unfortunately until they implement a For you algorithm across all instances and abandoned their archaic hashtag following system I will not be re-attempting