unsigned💀zero
@signedzero.bsky.social
Building software, breaking things, and reinventing wheels.
Art | Linux | Audio | Graphics | Games
Always learning.
Art | Linux | Audio | Graphics | Games
Always learning.
I think GH is financially incenstivised to do the opposite of this. Which is what they're doing now.
February 16, 2026 at 8:14 PM
I think GH is financially incenstivised to do the opposite of this. Which is what they're doing now.
> Are the test results made up?
This is a bigger problem than people realize. LLMs don't know or care if the tests actually passed.
If they "hallucinate" passing tests that's not even a "bug" really it's just the way they work.
This is a bigger problem than people realize. LLMs don't know or care if the tests actually passed.
If they "hallucinate" passing tests that's not even a "bug" really it's just the way they work.
February 16, 2026 at 7:57 PM
> Are the test results made up?
This is a bigger problem than people realize. LLMs don't know or care if the tests actually passed.
If they "hallucinate" passing tests that's not even a "bug" really it's just the way they work.
This is a bigger problem than people realize. LLMs don't know or care if the tests actually passed.
If they "hallucinate" passing tests that's not even a "bug" really it's just the way they work.
I'm really interested to see the way the open source community responds to this problem in the long run.
It seems undeniable that big changes are coming to the way we manage OSS projects.
It seems undeniable that big changes are coming to the way we manage OSS projects.
February 16, 2026 at 7:53 PM
I'm really interested to see the way the open source community responds to this problem in the long run.
It seems undeniable that big changes are coming to the way we manage OSS projects.
It seems undeniable that big changes are coming to the way we manage OSS projects.
Thankfully this is a slightly more reasonable approach at the very least.
February 16, 2026 at 7:51 PM
Thankfully this is a slightly more reasonable approach at the very least.
GH seems to be really leaning into selling themselves to project managers over developers these days.
February 16, 2026 at 7:50 PM
GH seems to be really leaning into selling themselves to project managers over developers these days.
What I'm also not really seeing is cases where LLMs actually work reliably enough to "let loose" like this. Even the frontier models.
Maybe I'm just working in a different problem domain or something?
Maybe I'm just working in a different problem domain or something?
February 16, 2026 at 7:48 PM
What I'm also not really seeing is cases where LLMs actually work reliably enough to "let loose" like this. Even the frontier models.
Maybe I'm just working in a different problem domain or something?
Maybe I'm just working in a different problem domain or something?
Step one to harnessing LLMs is to have a rational and educated discussion about how they work and what their limitations/risks are.
We have VERY clearly skipped that step, so a bit of overcorrection is in order IMO.
We have VERY clearly skipped that step, so a bit of overcorrection is in order IMO.
February 16, 2026 at 7:45 PM
Step one to harnessing LLMs is to have a rational and educated discussion about how they work and what their limitations/risks are.
We have VERY clearly skipped that step, so a bit of overcorrection is in order IMO.
We have VERY clearly skipped that step, so a bit of overcorrection is in order IMO.
I still really struggle to understand why people are letting an LLM make commits for them. When I see this, I assume they have not tested anything or even reviewed the code / ensured it matches the style guide etc.
February 16, 2026 at 7:37 PM
I still really struggle to understand why people are letting an LLM make commits for them. When I see this, I assume they have not tested anything or even reviewed the code / ensured it matches the style guide etc.
The more you look, the more it slops
February 16, 2026 at 7:31 PM
The more you look, the more it slops
Pretty sure people are not "ignoring" AI as much as they are seeing it, understanding it, and hating it.
February 16, 2026 at 7:30 PM
Pretty sure people are not "ignoring" AI as much as they are seeing it, understanding it, and hating it.
Check out TIC-80 as well. Very similar to Pico8 but totally free and open source.
February 16, 2026 at 7:55 AM
Check out TIC-80 as well. Very similar to Pico8 but totally free and open source.
Square fonts are a vibe for sure
February 16, 2026 at 5:43 AM
Square fonts are a vibe for sure
Doing basically anything "indie" these days from game dev to music is an exercise in getting and keeping peoples attention more than it's an exercise in the art form itself.
February 16, 2026 at 3:42 AM
Doing basically anything "indie" these days from game dev to music is an exercise in getting and keeping peoples attention more than it's an exercise in the art form itself.
Wow is this poly grid? I've never seen such a giant patch.
February 16, 2026 at 3:23 AM
Wow is this poly grid? I've never seen such a giant patch.
I'm honestly confused about why mastodon isn't more popular.
It completely solves the issue of who "owns" the platform at the very least.
It completely solves the issue of who "owns" the platform at the very least.
February 16, 2026 at 3:20 AM
I'm honestly confused about why mastodon isn't more popular.
It completely solves the issue of who "owns" the platform at the very least.
It completely solves the issue of who "owns" the platform at the very least.
These are fun because you can solve them without really memorising any algos.
But as you found out, kind of on the simpler side. Still very fun though!
But as you found out, kind of on the simpler side. Still very fun though!
February 16, 2026 at 12:27 AM
These are fun because you can solve them without really memorising any algos.
But as you found out, kind of on the simpler side. Still very fun though!
But as you found out, kind of on the simpler side. Still very fun though!
Scripting in fish is significantly more pleasant than bash IMO. The obvious downside though, is that fish isn't likely to be installed on a random nix system.
February 16, 2026 at 12:25 AM
Scripting in fish is significantly more pleasant than bash IMO. The obvious downside though, is that fish isn't likely to be installed on a random nix system.
The lack of a proper type system is one thing that gets brought up a lot when this is discussed. Duck typing is helpful until its not.
February 15, 2026 at 7:58 PM
The lack of a proper type system is one thing that gets brought up a lot when this is discussed. Duck typing is helpful until its not.
This is going to make some people very happy
February 15, 2026 at 7:18 PM
This is going to make some people very happy
As someone who has probably thousands of hours in the MC "beta" this post makes me feel so old.
February 15, 2026 at 7:03 PM
As someone who has probably thousands of hours in the MC "beta" this post makes me feel so old.
Neat. This seems like it could be a good middle ground between something like nano and a full featured editor/ide like neovim.
How was your experience working with Ratatui?
How was your experience working with Ratatui?
February 15, 2026 at 6:59 PM
Neat. This seems like it could be a good middle ground between something like nano and a full featured editor/ide like neovim.
How was your experience working with Ratatui?
How was your experience working with Ratatui?