It used to be more interesting and elaborate, switching between different models. I used to use Cursor, then Codex, but since Opus 5.5 came out, I just use the Claude IDE. I got rid of all my skills and just set it to medium. Then I vaguely talk to it about very high-end things that I want, and it comes back with magic. Rinse and repeat.

I’d say the biggest trick is just to tell it what complete looks like. Be very specific about what you want it to come back with once it’s done. That gives it a good idea what the goal is. It’ll automatically create subagents and all that stuff. Although if you have money to burn and you’re doing something large, you can tell it to just go wide or boil the ocean.

Also, make sure you’re working on at least five different things. Otherwise, you’re sitting around for one agent.

Hackernews comments always make me feel like I am robotripping and I accidentally crossed into a different reality. There are worse things than death…

Thanks for the response! Yeah it is a really nice convenience.

My point is couldn’t someone exploit that with a website that has a url matching a common grammatical error of this type?

You could put the chicken before the egg too by making a bunch of bot comments hat appear plausible except for errors like that and then actually launch the website right?

If I “accidentally” link to that hostile website by making a totally normal “grammatical error” how do you establish intent?

From a zoomed out database/admin level how do you quickly tell the difference between the two in bulk amounts of posts/comments you can’t possibly search all the way through by hand…?

Note here, in order for instances to properly federate in some sense copies of posts and interactions have to be made across multiple servers, this is one of the basic security/data control tradeoffs of the fediverse but the positive is that it naturally sets up the possibility (if that is what people want) for data to be “backed up” in multiple places making truly dead links a less likely outcome. A database engineer might look at the fediverse and scream about duplication of data all over the place but that is only one perspective to take on design, resiliency and redundancy have values we can easily invisibilize when we start quantifying everything.

The most important thing for making a fediverse that helps preserve human interactions I think is getting people excited about the generative capacity of sharing and duplication rather than just seeing these as a peculiarity of federation on ActivityPub and an inherent weakness/messiness to it.

Step 1. BURN AI AND GOOGLE DOWN TO THE GROUND WITH FIRE

Step 2. Hire human teachers and pay/treat them well

The only thing AI is actually good at is convincing unreflective centrists that humans are no longer very valuable, this works on the very humans you are aiming at devaluing too as is evident in this case.

Surely the reason kids are miserable is social media and not the world we have handed them!

Scam network temporarily pauses scam because of theft other news at 10

Why do we need to replace guide dogs?

Luddites specifically wanted machines to serve people, these people just hate everything.

Also I hope these people are getting paid well by Rupert Murdoch.

That video is immortal as

REVENUE NOT PROFIT