Mastodon Feed: Posts

Boosted by soatok@furry.engineer ("Soatok Dreamseeker"):
Firr wrote:

My computer just did a blue screen of death and, uh, this is new.

I wasn't watching discovery channel or anything on the computer I have no idea where that's from lol.

Boosted by jsonstein@masto.deoan.org ("Jeff Sonstein"):
jaykuo@universeodon.com ("Jay Kuo") wrote:

A fair point. Even if he’s still alive, if McConnell can’t do his job then he ought to step down. We need new laws and rules around no show lawmakers.

Boosted by jsonstein@masto.deoan.org ("Jeff Sonstein"):
ianvisits@mastodonapp.uk ("ianVisits") wrote:

Plans unveiled for 2.5-metre-wide tunnels to carry waste heat beneath south London

Two long tunnels could be dug beneath south London to carry surplus heat across the capital, creating a low-carbon heating network that could eventually warm around 300,000 homes.

https://www.ianvisits.co.uk/articles/plans-unveiled-for-2-5-metre-wide-tunnels-to-carry-waste-heat-beneath-south-london-91233/

dysfun@treehouse.systems ("gaytabase") wrote:

i did a really simple and nice thing that is almost certainly not the final answer i will settle on - i have a single pool per thread of spare individually allocated cachelines to reuse for three kinds of queue object. for a cross-thread message queue, it can hold one message. for a thread-local queue, it can hold two and for a gin-local queue it can hold 3.

it's the thread-local queue i don't think it's right for. you can only store 2 of them in a node because entries are 3 pointers in size. gin queues are likely to remain extremely short in practice (in fact empty for most gins).

Boosted by neatnik@social.lol ("Neatnik"):
peter@thepit.social ("Peter") wrote:

tech is pivoting to humanoid robots!!

the humanoid robots:

Attachments:

• video: 1cc4ca874f59f263.mp4

dysfun@treehouse.systems ("gaytabase") wrote:

just implemented receive and selective receive from the gin-local queue for my gin actors. yes they're basically actors now.

selective receive is about twice as long.

baldur@toot.cafe ("Baldur Bjarnason") wrote:

Thinking out loud for a bit:

The problem with tech, as others have pointed out, is that tech as a whole isn’t exactly a humane environment filled with thoughtful and charming people, and it’s in a multi year cycle of layoffs. So, if the job itself is losing what fun it had most other industries start to look appealing in comparison.

And if I manage to get out of the coding-for-money racket and end up missing bits of it, I could always take up coding as a hobby

Boosted by kornel ("Kornel"):
halcy@icosahedron.website ("halcy​ :icosahedron:") wrote:

boubas delivery service

Boosted by kornel ("Kornel"):
firefoxwebdevs ("Firefox for Web Developers") wrote:

popover="hint" landed in Firefox 153. Here's how it differs from regular popovers:

Attachments:

• video: 04e3453c8befb705.mp4

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

And yes, I do think there is a way to do this, and if I weren't deeply invested in the work I am doing, I might be working on it right now, but also wrestling with this very dilemma above.

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

A hypothetical: Imagine I held a magic wand that could make AI systems improve in all the following ways:

- More environmentally friendly
- Less capturable by a few large corporations
- Less built on "plagiarism machine" foundations
- More reasonable and capable of making judgements which are explainable, accountable, and more accurate
- More participatory and open

However, we are still left in our present socioeconomic system.

Should I wave the magic wand and improve things, or does improving AI in these ways still make the world worse?

Boosted by cwebber@social.coop ("Christine Lemmer-Webber"):
cwebber@social.coop ("Christine Lemmer-Webber") wrote:

@mntmn I talk a bit about it in this talk https://www.youtube.com/watch?v=jI8gA68OXLM

But really it's using machine learning models PLUS symbolic reasoning systems (prolog, propagators, etc), which can help make them more reasonable for cheaper.

Think of the human brain as having two modes of thinking: a fast, gut thinking approach, and a slower reasoning approach. All the machine learning model research has been about beating the former into as powerful a shape as possible by throwing tons of compute at it. But we also have knowledge of how to do symbolic reasoning. And we can make things a lot better by combining them.

Leilani Gilpin's research is very interesting in this regard https://people.ucsc.edu/~lgilpin/

In particular, I recommend her dissertation which asks: if a car drives off the side of the road, how do we hold it accountable? https://people.ucsc.edu/~lgilpin/publication/dissertation/

Boosted by dysfun@treehouse.systems ("gaytabase"):
gsuberland@chaos.social ("Graham Sutherland -> EMF 📞8087") wrote:

"EMF Camp obtained written permission from David Thorne of 27B/6 to use his classic spider drawing for their EMF Spiders shitposts" and other truly incredible sentences

pzmyers@freethought.online ("pzmyers 🕷") wrote:

Not gonna lie, I'm out to indoctrinate people.

https://freethoughtblogs.com/pharyngula/2026/07/22/welcome-to-my-den-of-iniquity/

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

Vibecoding gets banned on Codeberg! https://codeberg.org/Codeberg/org/pulls/1253#issuecomment-19820434

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

But it isn't now, and that's a whole future topic, and it still won't fix all the socioeconomic problems that LLM *corporations* are inserting into the world. But that's another topic.

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

That said, I do think there's a way to make machine learning models (not just LLMs, but including them) safer with neurosymbolic programming. And things are starting to trend in that direction but the industry is mostly just dumping all its research into beating large models to death rather than working on *systemic* improvements.

That said, I think it'll get there.

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

But now LLMs are themselves a source of vulnerabilities! That's a new vulnerability class! These companies are trying to *sell* their LLM tools for security while also telling you, put them everywhere!

And while they're great for finding exploits, *LLM agents ARE confused deputies*! They inherently are unable to prevent themselves from mingling the authority they are given. You can contain and sandbox them, but you cannot make *their behavior* safe, because they are gullibility machines.

So part of the irony for increased LLM proliferation making our world less safe is: STOP INSERTING AGENTS EVERYWHERE

They can be useful for some things particularly in the attacking domain, but for generating code, for understandability of systems, for places where you want an interface for a human being, they're not safe right now.

Though, there is a path to making LLMs safer too.

dysfun@treehouse.systems ("gaytabase") wrote:

re: LB: holy shit my scrollbars are fixed

Boosted by kornel ("Kornel"):
valkyrie@wandering.shop wrote:

very excited to announce the launch of my new business venture

https://overpaid.lol/

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

Most security vulnerabilities have pretty common shapes. But we also have solutions to them:

- Ambient authority! Mitigation: capability security
- Confused deputies! Mitigation: capability security
- Memory exploits: Mitigation: use memory safe languages (Rust and etc if you gotta go low level, though languages with a GC or etc are actually even *safer* because they don't have "unsafe" blocks of code) or use CHERI
- Injection attacks: Use structured representations. Quasiquote or combinator templating is much safer than string templating, for example. Why are we speaking to our databases in sentences? Send them the AST of the query!
- Bad crypto usage: Use standardized cryptography patterns

There are a few others, but even if you just focus on the above, we can simply *eliminate* huge swaths of security issues.

Boosted by dysfun@treehouse.systems ("gaytabase"):
dalias@hachyderm.io ("Cassandrich") wrote:

TIL in Firefox about:config:

widget.gtk.overlay-scrollbars.enabled = false
widget.non-native-theme.gtk.scrollbar.thumb-size = 1
widget.non-native-theme.scrollbar.style = 4
widget.non-native-theme.scrollbar.size.override = 16

Gets you scrollbars that are actually visible and clickable with the mouse.

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

I really don't think that you can fight LLM agent exposed and exploited security issues with *more vibes*. You absolutely cannot. These machines are good at finding exploits, but they are also good at *inserting* them, accidentally all over the place, and also in terms of them being the perfect machines to insert subtle, hard-to-find intentional vulnerabilities.

In terms of using them as vulnerability scanners, I think they're important tools. In terms of code generation, there is simply no way to fight vibes with more vibes.

Instead, we need to get systemic about things.

So how do we fix security issues SYSTEMICALLY?

cwebber@social.coop ("Christine Lemmer-Webber") wrote:

To those saying the OpenAI + Hugging Face thing is an ad, I think that's true, but I think it's not just an ad. They're taking the marketing opportunity for sure, and it's *awfully convenient* that it's two AI companies as part of the story, right?

But, despite being a big critic of this stuff, I do think that they are *very good at attacking systems*. The reason being that most attacks tend to have very similar patterns, and these models are great at blasting similar patterns over and over again, so they can ruthlessly find their way through exploits.

The bigger question to me is: which examples *aren't* there motivations for two companies to put out statements together? Where else has stuff like this happened?

And the more important one, the response I think they *want* people to hear is "our stuff is so scary and powerful, you have to use and trust even more of our vibetech".

But that's not MY takeaway...

Boosted by brib@bribstodon.xyz ("brib :neofox_floof:​ :Nonbinary:"):
Blort@social.tchncs.de ("Blort™ 🐀Ⓥ🥋☣️") wrote:

@ms_zwiebel

I haven't settled yet on what's a decent VPN alternative, although my current frontrunners are Njalla (run by a co-founder of The Pirate Bay) and Nebula by Defined.net. Not sure about their politics, but distributed and Nebula is FOSS which is a positive signal, while not a guarantee.

What I've ruled out:
ProtonVPN - Actively working with government against activists
Windscribe - Transphobic indications
Mullvad - Funding a fascist leaning party

I'd love to hear anything people have learned about the alternatives, either pro or against. Boosts appreciated for reach.

#Mullvad #VPN #Privacy #Politics #ProtonVPN #Windscribe #Njala #Nebula #DefinedNet

baldur@toot.cafe ("Baldur Bjarnason") wrote:

Just realised that 2026 is the twentieth anniversary of me finishing my PhD and leaving academia for good.

Boosted by cwebber@social.coop ("Christine Lemmer-Webber"):
shauna@social.coop ("Shauna") wrote:

@cwebber

I feel like OpenAI is going for "our models are so smart!" but what it's giving is "our sandboxing is so incompetent!"

brib@bribstodon.xyz ("brib :neofox_floof:​ :Nonbinary:") wrote:

The disadvantage: Christmas stuff usually stays in a cupboard 91.7% of the year.

I don't really want to solder my Arduino to the Christmas Tree kit if I can avoid it, as that feels like a waste of a good Arduino.

(Perhaps I can get one of those super cheap microcontrollers for the kit that I don't mind sticking in a cupboard so long?)

dysfun@treehouse.systems ("gaytabase") wrote:

Boosted by brib@bribstodon.xyz ("brib :neofox_floof:​ :Nonbinary:"):
gedankenstuecke@scholar.social ("Bastian Greshake Tzovaras") wrote:

I co-proposed/wrote the #Codeberg vibe-coding ban and all I got was a lot of haters in my mentions.

And a nearly 70% approval rate, with one of the highest voter turnouts the association had afaik (and one that's comparable to the last EU elections). 😎