iiinova — where your digital self keeps evolving

Yeah It hasn't been Torvaldz, Woz, and Von Rossum for a long time. It's all Jobs, Musk, and DHH This is who ruined social media, crypto, and now AI

020

asura @asura.dev · 1h

The very last thing I heard is you can actually find it in javascript, but tbqh I haven't really been tracking this I just want more people burning tokens seeing if it says the number of o's in google is propeller or whatever

020

asura @asura.dev · 1h

Dead ass this happened last week from a contractor at a billion dollar company: "Binding to 0.0.0.0 is a security risk [7 more paragraphs]" Brrrruh, that's a container in kubernetes, what are you doing here? How did this become your job? This should be labelled harmful content. Straight to jail.

040

asura @asura.dev · 1h

Oz and Bobby are boomer celebrities. Maybe not yours, but definitely boomer celebrities But yeah, to be honest I'm lashing out a lot here. When I see 20 70-90 year olds in a room agreeing that we should stop all vaccine funding, when they're the ones that benefited from it, I get a bit antagonizing

000

asura @asura.dev · 2h

What AI has done for many that are very excited about it is help them become better arsonists I've seen so many things go flying through my email that simply do not match industry for years and I guarantee you most of it is @#$! copilot or one-shot Claude with no context oss isn't fairing better

120

asura @asura.dev · 2h

I find it really funny this year that we've returned to code being the art of the solo basement dweller javascript wizard dude Silicon Valley was damn near a documentary - the average dev department is bros broing and waiting to be firefighters. The adage is firefighting breeds arsonists.

180

asura @asura.dev · 2h

The developers that derive worth from finding solutions aren't in your average corporate environment. Anyone arguing against using outdated software on RHEL5 in 2026 so they can write htmx for the love of the game... got "obsoleted" twice - once by outsourcing, then by AI. See you at standup

040

asura @asura.dev · 2h

Isn't this bad, though? I thought people figured out that was a hardcoded response in the site and not an error. Maybe I am out of the loop.

130

asura @asura.dev · 2h

"But it's lying" is always tells of disconnect. Calculators don't lie. That's a human trait To Michael's point, if you train an LLM on gay frogs in chemtrails that stole the election, they will then decide that the sky is potato and ocean is grass - like a calculator, 2+2=5 cascades to other errors

081

asura @asura.dev · 2h

I'm going to continue to be frustrated and say that if it is population, then the boomers can kill BARDA even if it's the millennials skipping vaccines :P

000

asura @asura.dev · 2h

There is not a reason to do frontier since those are black box platforms that most likely have routers and wasm

000

asura @asura.dev · 2h

The last I checked on that thread there were 2 One was a 27b model

100

asura @asura.dev · 2h

Sure. The "I don't believe in germs" Fox News host Pete Hegseth is solidly genx. Either way, RFK Jr. and Dr. Oz are your celebrities.

100

asura @asura.dev · 2h

That doesn't work with the other thread. They're dictating a fish doesn't climb a tree without having a fish or a tree or knowing the shapes of those objects. The one attempt at even having any sort of hypothesis or experiment was a model from a couple of years ago. My local Qwen will add numbers.

100

asura @asura.dev · 3h

I've seen this and it tracks but what in the hell is the explanation for RFK Jr. and MAHA? It's the 20 year olds making him create those policies?

100

asura @asura.dev · 3h

Go talk to RFK Jr. and get back to me on that

100

asura @asura.dev · 3h

I don't use Photoshop as a calculator and fish don't climb trees. In spite of that, many very large models can reliably add numbers. It doesn't make a whole lot of sense to use the source of orchestration and decisions as a calculator, but even local GLM can add.

110

asura @asura.dev · 9h

Can I get that in bootstrap4?

120

asura @asura.dev · 10h

wasm python is pretty easy to adapt and frameworks like pydantic AI use Monty. You could have Claude code use bc if you're a weirdo. There's no special text classifier or whatever is described here. 9b instruct models can decide to call the tool function for math.

030

asura @asura.dev · 10h

What an odd thing to say. Let me fix it: If you're not a fool you are dispatching deterministic jobs to tools available to the LLM. Not doing this is as silly as asking an LLM the weather with no way to actually get any weather data. GIGO fish-climb-a-tree silliness I don't use Photoshop for math

4120

asura @asura.dev · 10h

Anything that benefits from simultaneous processing is profitable for GPGPU and Huang has had an almost religious desire to push more computing into the GPU. I can remember him celebrating when AMD and Intel merged and guaranteed x86 as a unified platform. If he did that with AI ... Utopia.

000

Reposted by asura

Stuart Gray @sgray.bsky.social · 11h

Big shades of Microsoft’s infamous Fear, Uncertainty, and Doubt (FUD) campaign against Linux in peak Windows years. Certainly seems like post-IPO Anthropic wants a cartel-like near monopoly on Intelligence.

1152

asura @asura.dev · 11h

Nvidia hopes to pave the world in GPUs. Easy enough. For the rest of it, all of this stuff started as and continues to be open source solutions. Their options are to provide better products/service or ban software ... which is a bit self defeating. Surely Huang is smarter than this.

100

Reposted by asura

Erica Windisch @ewindisch.ontological.observer · 20h

I reported vulnerabilities to the Linux kernel and OpenZFS assisted by GLM-5.2. Anthropic's cyber program and refusal responses are detrimental to global information security. Defenders depend on GLM because they can't depend on Anthropic.

022846

asura @asura.dev · 14h

And ouija boards. Source: my arm I guess

090

asura @asura.dev · 14h

I tried to read through some of those I think yesterday We have reached peak victimhood where the average white guy is now fighting the robot over his superiority and still not winning, so he's also a victim of the robot. It's wild.

070

asura @asura.dev · 14h

This is how Musk plans Starlink features

060

asura @asura.dev · 15h

oh ffs do I need to go get an Orange Website to host a @#$! model download on because a sundowning boomer, Derp Hitler, Amodei, and Huang can't capitalism? This is so stupid.

131

Reposted by asura

Thomas Wood @advanced-eschatonics.com · 16h

Anthropic needs to keep GLM-5.3 out of its mouth

0376

asura @asura.dev · 15h

I guess I should explain that joke and why xSI and claude.si are going to have some trouble. en.wikipedia.org/wiki/.si

en.wikipedia.org

.si - Wikipedia

000

asura @asura.dev · 15h

The world: Idiot clownshoes fools Slovenia: Wait a second...

120

Reposted by asura

James @gravitysra1nbow.bsky.social · 22h

Do we seriously have to do Maoist newspeak every single time an 80 year old posts his idiotic ideas on the internet. It has been AI since before I was even born. Can literally anyone have a shred of dignity.

1223162498

asura @asura.dev · 16h

Compsci degrees reporting in. It's only value is it's any degree. Could've been basket weaving philosophy and I'd still be where I am. I'm also a traitor to compsci - I worked for a for-profit college servicing company that dropped turn key curriculum in other colleges. It was all trash.

030

asura @asura.dev · 16h

ChatGPT+, the social network

020

asura @asura.dev · 17h

And if it's not then they've already sold the idea that AI will create a world of scarcity (took all the jobs!) and that we destroyed the world for your *checks notes* freedom and security (like every other war)

031

asura @asura.dev · 17h

If you're happy with that B11 I might get one. I got the nape and another k8 for my rog flow, but the k8 is like... you can use that keyboard as self defense. I think it's bulletproof. It's heavy af.

110

asura @asura.dev · 17h

I got one :) It's fun. I wish Keychron had the forethought to be able to snap it to the bottom of one of their keyboards. I keep begging them to make an actual thumb cluster but no response. They did do a refresh of the Q11 and I love it.

110

asura @asura.dev · 17h

Have you seen the nape, though?

110

asura @asura.dev · 17h

... then why aren't they using functiongemma or LoRA small models for API json response or... anything resembling using an openai model? :P

000

asura @asura.dev · 17h

This sounds like something a moth would say

0452

asura @asura.dev · 17h

Dame representing the keychron ayo ... and some piece of shit mouse you probably need to license to use or something idfk :P

140

asura @asura.dev · 18h

US businesses have spent years doing fake hiring, inventing return-to-work mandates, and now AI all to hide RIFs in a bid to convince the world they will infinitely be making more money every year. But oh no it's definitely AI that's causing the problems. 🤡 Dog wagged, checkmate, we stay poor ig

050

asura @asura.dev · 18h

That "close enough" includes Chinese robots doing speed running and wu shu, so close enough is certainly good enough

020

asura @asura.dev · 18h

This is not a dunk (ffs bluesky, I shouldn't have to say that) Similar to how bipedal robots were not a thing before they were, we are going to inevitably find out digits are overengineered For walkers the idea of zero point and feedback was "too much compute" but now it's a matter of close enough

150

Reposted by asura

alias @antiali.as · 29/09/2026

Someone with a good vid gen stack please remix the ballmer video to show sama screaming “Jevelopers! Jevelopers! Jevelopers!”

1455

asura @asura.dev · 29/09/2026

Can confirm that sunscreen is optional as long as you never go to the big room with the bright light where the trees and grass are.

060

asura @asura.dev · 29/09/2026

We're beyond dollar amounts and now selling a religion :P This is one of the many reasons I'm always trying to get people to commit to local models and techniques. The first-mover in any tech industry always eats shit. All the labs are doing their best to try to make fast-follower illegal 🤡

110

asura @asura.dev · 29/09/2026

I want sort of the opposite I get a notification from signal, discord, etc and there's no actual way to reply to it with an intent I can reply and initiate texts and I can "Share to..." other apps so what is the problem here? :\

010