Ember AfterDark

@ThePrimeagen wait does anyone know what killer features github is adding for the sake of all this downtime? or maybe its a psyop to kill the platform?

dont you do it github...

2026 year of instability

@ThePrimeagen 2026 is the year of instability... ai being right most of the time leads to complacency amongst bad developers because who has time to check all those lines, most of them are probably right...

What if they bring back Jack O'Neill's clone in the new Stargate series? @GateWorld

Playing around with Unreal Engine youtube.com/live/5lYaUDJ73… via @YouTube

The problem with a tool that works most of the time is humans are lazy.

Thumbs Up 👎

Developers training AI

before you make an LLM do something... ask yourself if this task can be completed using a deterministic algorithm Why? because you are trading a convenience now for a nightmare tomorrow instead just use LLM to create a deterministic solution

I think it struggles to simplify projects because you are asking it to do something entirely new and the options for an entirely new pattern are endless... so it just picks one.

LLMs seem to follow an old standard of garbage in, garbage out. If your project has bad patterns, AI will gladly duplicate those patterns to eternity and when it is time to 'reduce/simplify' ... LLM will struggle to maintain the same result with a simpler solution.

@theo funny enough, I'm making actively developing a new linux distro... Or should I say a linux non-distro. Well this pretty much seals the deal... I won't distribute my solution, problem solved...

Linux might become illegal in California. I wish I was joking.

@thdxr I was just thinking about this. AI giving good responses most of the time is problematic. It can make you complacent and relaxed when reviewing code or cause some to just stop reading all together. We still need to read code. 2026 is the year of instability.

mfs will do anything but read the code

@elonmusk Some how this is not even close to the truth. Reality is much worst.

Giving people agentic AI be like …

@trq212 nice idea but why not just open a second terminal and keep things organized?

We just added /btw to Claude Code! Use it to have side chain conversations while Claude is working.

@ThePrimeagen But I did get to take them so that is cool.

"you miss 100% of the shots you don't take" With AI we learned that you also miss 100% of the shots you do take because your ideas are trash

Linux moving from X11 to Wayland is a massive security upgrade. X11 basically let any background app spy on your screen and keystrokes. Wayland fixes this by locking apps into isolated "surfaces" so they can't see each other. The downside is old apps running through the XWayland compatibility layer still think they own the whole monitor. That clashing logic is what causes weird visual bugs, like Steam menus spawning in the wrong place. Your experience really just depends on whether your specific compositor uses hacky band-aids to fix those old apps, or actually built a solid XWayland implementation. Anyways, I'm building a Wayland compositor, just discovered Unreal Editor relies on X11, and now my project scope has exploded into adding a native Wayland support to Unreal because I am too invested to stop now. Has anyone else got Unreal Editor Wayland stable yet? @UnrealEngine on #wayland

Goals should always guide tasks. If you only delegate tasks, the result may miss the target. But if you delegate the goal along with the tasks, the process can change, but the goal gets accomplished. Lead AI with goals.

@ThePrimeagen I'm just going to build an alternative.

asking for a friend... github, could you please have 2 9s of reliability?